Development and Simulation of an Innovative Autonomous Knowledge-Based Smart Waste Collection System

Development and Simulation of an Innovative Autonomous Knowledge-Based Smart Waste Collection System | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Development and Simulation of an Innovative Autonomous Knowledge-Based Smart Waste Collection System Mohamed Abdallah, Mariam Hosny This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8012685/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 26 Apr, 2026 Read the published version in Scientific Reports → Version 1 posted 15 You are reading this latest preprint version Abstract Substantial efforts have been directed to modernize and decarbonize the waste collection industry. Smart sensor-based waste collection (SWC) systems were developed to optimize collection routes based on the actual waste levels in bins, resulting in reduced service frequency, fuel consumption, and air pollution. To date, there has not been a fully functional commercial SWC system due to the complexity and limited practicality of such a hardware-intensive design. Alternatively, this research presents an innovative cloud-based approach in which historical waste generation data, acquired from onboard-truck sensors, replaces data from bin sensors. The proposed knowledge-based waste collection (KWC) system incorporates machine learning for waste forecasting, expert bin selection, route optimization, and autonomous navigation. The system was simulated in an actual residential district to collect recyclables over three scenarios: conventional, SWC, and KWC. Historical data were used to train a generalized linear model (GLM) and a gradient-boosted tree (XGBoost) model to predict daily waste generation per bin. The heuristic bin-selection algorithm selected the bins to be served based on the actual and forecasted waste quantities in SWC and KWC, respectively. XGBoost achieved higher prediction accuracy than GLM with a 4.2% relative error. SWC and KWC significantly reduced the travel distance by 63.1% and 60.9%, respectively, whereas the number of collected bins decreased by 89% for both scenarios. The number of collection days decreased by 5 and 3 days per month in SWC and KWC, respectively. The implementation of connected and autonomous vehicles (CAVs) significantly improved the system, decreasing the total delay by 91% and 90% in SWC and KWC, respectively. Moreover, a life cycle costing analysis revealed that, compared to conventional collection, the reduced travel expenses of SWC were insufficient to offset the cost of bin sensors, whereas KWC achieved a 63% cost reduction by replacing hardware-intensive components with a cloud-based system. Overall, this research has demonstrated that KWC systems can potentially outperform hardware-intensive SWC systems, given the substantial economic and operational benefits associated with such a cloud-based approach. Physical sciences/Energy science and technology Physical sciences/Engineering Earth and environmental sciences/Environmental sciences Physical sciences/Mathematics and computing Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 Figure 12 Figure 13 1. Introduction Solid waste management is one of the major challenges faced by municipalities worldwide due to its direct impact on the environment and public health, as well as its economic burden and extensive operation. The rapid growth in waste generation rates has led to more challenging management processes, particularly in the collection system. The conventional waste collection approach involves a fleet of trucks transporting waste collected from bins to intermediate or final disposal sites. This process leads to ongoing issues, including air pollution and traffic congestion, and typically accounts for more than half the total waste management costs. Moreover, the daily and seasonal variations in waste generation rates result in low efficiency of the traditional predefined fixed collection route regardless of the actual bin use capacity (Abdallah et al., 2019 ). This results in the frequent servicing of waste bins that are only partially full at the time of collection (Ramos et al., 2018 ). The socioeconomic and environmental implications of such inefficient operation suggest that smart collection systems that respond to the actual bin use will provide potential improvements. Recent efforts have focused on modernizing this inefficient process by developing sensor-based smart waste collection systems, aiming to reduce the frequency of waste collection events and consequently lower fuel consumption, carbon emissions, and traffic congestion. Smart waste collection can be categorized into two areas: 1) data acquisition and communication hardware, particularly at the waste bin, and 2) decision-making and route optimization algorithms, along with simulations under various operating scenarios. Overall, compared to conventional systems, previous studies reported theoretical improvements in operation and maintenance costs, total travel time, fuel consumption, and air pollution emissions. Most hardware studies focused on installing sensors in waste bins (Mamun, Hannan, Hussain, & Basri, 2016); (Gutierrez, Jensen, Henius, & Riaz, 2015 ); (Catania & Ventura, 2014 ); (Juwariyah et al., 2021 ); (Sidhu et al., 2021 ), whereas studies that considered onboard truck sensors are lacking. However, several challenges limit the practical feasibility and scalability of these systems, including heavy reliance on sensor instrumentation, periodic maintenance, significant upfront costs, calibration, signal interference, deliberate or unintentional tampering, and power supply issues, among others. Collection trucks equipped with onboard weight sensors may offer several advantages compared to sensor-equipped bins, such as 1) elimination of the required wireless communications between bins and a control center, 2) fewer monitored vehicles compared to numerous waste bins, 3) easier centralized maintenance at the garage versus scattered service points (bins), 4) longer sensor lifetime due to less indirect contact with waste and continuous extreme weather conditions, and 5) security of the installed sensors due to inaccessibility to the public and protection against vandalism, all of which lead to a more robust, practical, simple, and cost-effective operation. This study replaces sensor-equipped bins with weight sensors onboard trucks that record the amount of daily generated waste at the bin level. Additionally, the amount of waste per bin can be forecasted for the next day using the recorded quantities, allowing for optimized collection routes for bins that require servicing. 2. Literature Review The developed knowledge-based waste collection system integrates daily waste generation forecasting, bin-selection decision-making, route optimization, and autonomous transportation. The following section provides a comprehensive literature review, examining each component as a standalone technology and as an integral part of a dynamic waste collection system. 2.1. Waste Generation Forecasting Several studies have forecasted waste generation with varying temporal and spatial resolutions, where most of the literature is focused on predicting the annual (Jassim et al., 2022 ; Lu et al., 2022 ; Latif et al., 2023 ; Srivastava & Jha, 2023 ) and monthly (Abbasi et al., 2019 ; Fan et al., 2021 ; Ghanbari et al., 2021 ; Liang et al., 2021 ) produced quantities in cities and countries. Multiple studies have predicted the amount of solid waste generated in cities, varying the temporal resolution from weekly to daily, as summarized in Table 1 . For instance, the weekly household waste (HSW) in nine regions Southeast of the United Kingdom was forecasted using a proposed ensemble meta regressor, which applied linear regression (LR) as a meta learner to be trained on the predicted optimized outcome of various machine learning models (Namoun et al., 2022 ). The meta regressor outperformed all the individual models, achieving a coefficient of determination (R²) of 0.80. The weekly refuse and recyclable waste were forecasted in 609 sub-sections of New York City using a Gradient-boosting regression tree, which achieved an R 2 of 0.87 (Kontokosta et al., 2018 ). Table 1 Summary of the best-performing machine learning models forecasting waste generation in the literature. Reference Model Temporal Resolution Spatial Resolution Waste Type Factors R 2 Guha et al. ( 2023 ) Multiple linear regression Average daily 381 households in 31 wards in Rangpur, Bangladesh HSW 5 0.66 Adusei et al. ( 2022 ) Recurrent neural network long short-term memory Daily (2013–2021) City of Regina Landfill, Canada MSW 1 0.72–0.86 Mudannayake et al. ( 2022 ) Random forest Daily (2011–2013) Ballart, Australia (2001–2011) Austin, USA (2005–2019) Boralesgamuwa, Sir Lanka (2012–2018) Dehiwala, Sir Lanka (2012–2015) Moratuwa, Sir Lanka (2015–2018) MSW Timeseries MAPE: Ballart: 8.3% Austin: 9.0% Boralesgamuwa: 29.5% Dehiwala: 36.7% Moratuwa: 33.1% Namoun et al. ( 2022 ) Ensemble meta regressor Weekly (2011–2021) 9 regions Southeast UK HSW 1 0.80 Araiza-Aguilar et al. ( 2020 ) Multiple linear regression Average daily 118 localities in Cuenca del Cañón del Sumidero, Mexico MSW 2 0.975 Noufal et al. ( 2020 ) Multiple linear regression Per capita average daily (2 weeks in Jul 2017, Aug 2018, Jan & Feb 2019) 300 households in Homs, Syria HSW 5 Per capita: 0.56 Recyclable: 0.71 Organics: 0.84 Kontokosta et al. ( 2018 ) Gradient boosting regression tree Weekly (2013–2016) 609 sub-sections in New York City, USA Refuse and recyclables 31 0.87 Hoang et al. ( 2017 ) Bayesian model average method Average daily (2 weeks in 2015) 286 households in Hoi An, Vietnam HSW 5 0.33 Song et al. ( 2014 ) Simulated annealing-based variable weighted combining the chaotic model, ANN, and PLS-SVM Daily (2011–2013) Seattle, USA MSW Timeseries 0.94 This study Generalized linear model and gradient-boosting trees Daily (2019–2020) Abu Dhabi, UAE Recyclables 0.94 ANN: artificial neural networks, PLS-SVM: partial least square support vector machine, HSW: household solid waste, MSW: municipal solid waste, MAPE: mean absolute percentage error On the other hand, the daily municipal solid waste (MSW) generation was forecasted in five cities across Australia, the USA, and Sri Lanka, spanning different temporal durations (Mudannayake et al., 2022 ). Various timeseries and deep learning models were utilized based on single- and multi-model prediction approaches. Random forest showed the best performance with a mean absolute percentage error ranging from 8.3% to 36.7% using the single model forecasting method. Song et al. ( 2014 ) predicted daily MSW using a timeseries simulated annealing hybrid model, which combines and weighs the outputs of various models. The proposed hybrid model had the best performance, with an R 2 of 0.94 (Song et al., 2014 ). In addition, the daily MSW was forecasted by a recurrent neural network long short-term memory model in Saskatchewan, Canada, with an R 2 ranging from 0.72 to 0.86 for different meteorological seasons (Adusei et al., 2022 ). Other studies have focused on predicting daily waste generation at varying spatial resolutions, ranging from households to divisions within a city. The Bayesian model average method was employed to predict the average HSW, surveyed across two weeks from 286 households in Hoi An, Vietnam (Hoang et al., 2017 ). In addition, multiple linear regression (MLR) was implemented by Guha et al. ( 2023 ) and Araiza-Aguilar et al. ( 2020 ), in which the former predicted the average HSW for 381 households across 31 wards in Rangpur City Corporation, Bangladesh, to identify suitable landfill sites, whereas the latter forecasted the average daily MSW for 118 localities in Cuenca del Cañón del Sumidero, Mexico, with an R 2 of 0.975 (Araiza-Aguilar et al., 2020 ; Guha et al., 2023 ). MLR was also utilized to forecast the per capita average daily HSW, as well as the daily fraction of recyclables and organics for 300 households in Homs, Syria, where the prediction of organic waste achieved the highest R 2 of 0.84 (Noufal et al., 2020 ). 2.2. Route Optimization Most of the literature applied route optimization for waste collection at garbage accumulation points (Mahéo et al., 2023 ), dumping sites (Rambandara et al., 2022 ), central collection points/gather sites (Amal et al., 2018 ; Hu et al., 2023 ; Zhang et al., 2022 ), separation centers (Mohammadi et al., 2023 ), and districts (Greco et al., 2023 ). This is due to the adapted fixed routes in local neighborhoods, which are optimized in terms of travel distance and time. However, dynamic routing systems should continuously adapt due to daily variations in the amount of waste per bin, requiring ongoing optimization to efficiently service fluctuating needs (Alsobky et al., 2023 ; Mohammadi et al., 2023 ). Some studies have optimized the MSW collection route at the bin level using various methods. For instance, seven scenarios were proposed to optimize the travel distance and collection time for a neighborhood in Cairo, Egypt, by applying an integer genetic algorithm to solve the traditional salesman problem (Alsobky et al., 2023 ). The baseline collection scenario consisted of 235 bins with a volume of 240 liters of unsorted MSW serviced once daily by medium- and five small pick-up vehicles with manual loading and no compaction. In contrast, the best-performing scenario involved a medium truck with mechanical loading and compaction, collecting sorted waste from 73 collection points through two daily rounds, which reduced the travel distance and collection time by 70% and 57%, respectively. Another study developed a hybrid particle swarm optimization (PSO) and firefly algorithm (FA) using local search (I-HFPSO) to optimize the waste collection route of 280 containers in a district in Sanlıurfa, Turkey (Kaya, 2023 ). I-HFPSO outperformed linear programming, ArcGIS, FA, PSO, and HFPSO, resulting in a 31.1% reduction in travel distance compared to the current practice. 2.3. Bin-Selection Decision-Making Few studies have implemented dynamic waste collection, which accounts for the varying daily amounts of waste and optimizes collection routes accordingly, as summarized in Table 2 . For instance, a dynamic waste collection algorithm was proposed that employs a look-ahead heuristic to select bins that are full on the current collection day and those expected to reach capacity on the next collection day, which may not necessarily align with the subsequent calendar day, based on statistically estimated daily waste accumulation rates per bin (Jorge et al., 2022 ). A simulated annealing (SA)/neighborhood search (NS) algorithm determines additional profitable bins to collect and optimizes the collection route of the selected bins. The hybrid metaheuristic was compared to actual operations in Portugal, where different runs yielded a profit ranging between 45% and 60%. Additionally, an algorithm for bin selection and route optimization was developed using a mixed-integer linear programming model and tested on both small, randomly generated instances and large instances adapted from an existing split delivery vehicle routing problem (Luo et al., 2024 ). The developed system utilizes the variable NS framework and employs SA as the acceptance criterion. Sites for waste collection were selected based on whether they exceeded a set threshold or were predicted to overflow the next day. The system allowed multiple vehicles to service a site if its waste surpassed the capacity of one vehicle. The performance of the proposed model using small-scale instances was compared to four competing state-of-the-art algorithms and CPLEX, showing a reduction in operating costs and fleet size by approximately 50% when implementing an 80% collection threshold, effectively preventing overflow. Moreover, a multi-agent simulation-based modeling approach was proposed, comprising agents representing households, planners, and drivers (Hussain et al., 2024 ). Waste density was determined in real-time by wireless sensors that transmit data to a cloud platform. This data was then monitored by the planner agent, who decides whether to initiate a waste collection trip based on the number of bins exceeding a predetermined threshold. The wireless sensor-based scenario achieved significantly reduced travel distances, carbon dioxide emissions, collection times, the number of overfilled bins, and the number of trucks. Furthermore, a knowledge-based dynamic waste collection system was developed for a residential area (Abdallah et al., 2019 ). The economic and environmental benefits were estimated to reach 40% through analysis of waste generation data, a bin-selection algorithm, and geographic information system (GIS) route optimization. A smart system entirely dependent on hardware was considered impractical under the harsh and aggressive operating conditions of the costly waste collection process. Table 2 Summary of studies that implemented dynamic waste collection systems, including bin selection and route optimization. Authors Data source Waste type Monitored factors Monitoring method Bin selection conditions Route optimization constraints Objective function Bin capacity Capacity threshold Fleet size capacity (Luo et al., 2024 ) Simulated data for community collection sites - Level Waste generation quantity Initially known (unspecified) > threshold (collection day) > bin capacity (next day) < vehicle capacity Multiple vehicles can collect Min cost Small instances: 0–2Q Large instances: 0 - (1.5×I0i) Small instances: 0%, 60%, 100% Large instances: 0, 60% − 100% 2–41 (Jorge et al., 2022 ) Simulated data Real data (Lisbon, Portugal) Paper Cardboard Level Accumulation rate Real-time waste Level sensors ≥ bin capacity (collection day) ≥ bin capacity (next collection day) < vehicle capacity Maximum labor shift duration Maximum route workload balance Removal of arc intersections Model running time ≤ 2 hours Max profit 74 kg 100% 1–2** 4 tons (Hussain et al., 2024 ) Simulated data (Al Rayyan, Qatar) Household Density Real-time sensors > threshold > bin capacity < vehicle capacity Each bin collected by one vehicle 1 vehicle can collect multiple rounds Min distance Min cost 1 m 3 312 kg/m 3 60%, 70%, 80%, 90% - (Martikkala et al., 2023 ) Simulated data based on sensors (Seinajoki, Finland) Textile Level Real-time sensors > threshold Capacity/time constraint Min cost 0.24 m 3 90% (previous evening) 80% (collection day) 1 (Belhiah et al., 2024 ) Real data (Tangier, Morocco) Green - Real-time sensors - - Min distance - - 300 manual sweepers (Brouwer et al., 2023 ) Real data (ERSUC, Coimbra, Portugal) 2 Paper 1 Plastic/metal 1 Glass Level Real-time sensors Visual observation Mobile sensors - - - 2.5 m 3 - - This study Real data (Abu Dhabi, UAE) Recyclables Weight Predicted > threshold (collection day) - Min distance 90% 1 **Real data scenario, I0i: the initial amount of waste at customer 𝑖 (tons), Q: Vehicle capacity, Recyclables: paper, plastic, metal, and glass, UAE: United Arab Emirates Table 2 continued Authors Route optimization number of days Number of bins Utilized methods Optimization problem Scenarios BAU Collection Frequency Luo et al. ( 2024 ) 7 5–10 MILR VNS SA VRP SDVRP Variable capacity, VRPHAS, SRC + VND, SplitILS, RGTS, CPLEX Daily 5 27–43 Variable capacity Jorge et al. ( 2022 ) 5–10* 30** 50–250* 226** SA NS VRPP Exact branch-and-cut algorithm*** SANS without workload concerns* SANS with workload concerns* Actual operations** Hybrid metaheuristic** Variable Hussain et al. ( 2024 ) 365 200–500 FSM MTSP Periodic review (BAU) Wireless sensors-based Daily, after-2-days, after-3-days, weekly Martikkala et al. ( 2023 ) 28 10 ODL - Conventional (BAU) Dynamic Weekly (all bins) Belhiah et al. ( 2024 ) - - Contraction Hierarchies Open Source Routing Machine - - - Brouwer et al. ( 2023 ) - 4 - - - This study 30 396 ArcMap TSP Conventional (BAU) Sensor-based collection (real-time) knowledge-based collection (forecasted) Daily *Simulated data scenario, **Real data scenario, ***Developed by Ramos et al., ( 2018 ), BAU: Business as usual, FSM: finite state machine, MILR: mixed integer linear programming, NS: neighborhood search, ODL: Open door logistics, RGTS: randomized granular tabu search, SA: simulated annealing, SRC + VND: Split-delivery route construction improved by variable neighborhood descent, TSP: traveling salesman problem, MTSP: multiple TSP, VNS: variable NS, VRP: vehicle routing problem, SDVRP: split delivery VRP, SplitILS: SDVRP solved by iterated local search, VRPHAS: VRP heuristic with a priori split strategies, VRPP: VRP with profits 2.4. Autonomous Vehicles Researchers previously introduced the idea of incorporating connected and autonomous vehicles (CAVs) into a waste collection system (Alfeo et al., 2019 ). Autonomous and connected transportation can inform vehicles in advance about surrounding traffic and maneuvering decisions, thereby maintaining uniform speeds and minimizing sudden braking, which leads to reductions in fuel consumption, traffic congestion, and emissions. CAVs use a global positioning system (GPS) and light detection and ranging (LiDAR) to map, position, and scan the area around the vehicle. For waste collection applications, predetermined routes will significantly reduce the complexity of the driving task assigned to CAVs. This concept is crucial for determining the minimum level of automation required to perform the task. According to the Society of Automotive Engineers (SAE), all vehicles fall under one of six levels of automation, ranging from no automation (level 0) to full automation capabilities (level 5) (SAE, 2018 ). Over the last decade, researchers have tested and validated the reliability of Level 3 CAVs, with most car manufacturers having reached the development stage for Level 4 automation. Additionally, to receive signals from waste bin sensors and communicate with other collection vehicles, CAVs will also be required to have vehicle-to‐vehicle (V2V) and vehicle‐to‐infrastructure (V2I) capabilities. In 2018, the Volvo Group developed autonomous waste collection vehicles with V2V capabilities and demonstrated the potential benefits of utilizing these CAVs in navigating obstacles and consistently driving to each bin location. Other automotive companies have also deployed autonomous waste collection trucks, such as eCanter SensorCollect by Mitsubishi Fuso, in Japan, Europe, and the USA. However, the current research lacks the field application of autonomous vehicles for waste collection. The focus is mainly on three domains: 1) robots for collecting waste in water bodies, 2) robots for collecting trash from beaches and parks, and 3) simulation of autonomous vehicles for waste collection. For instance, a mobile autonomous robot for litter emptying (MARBLE) was developed and tested on three different areas with varying numbers of bins (Justo et al., 2023 ). Routes were optimized using vehicle route problems with SA, game theory knapsack problems with smart litter bins, and knapsack problems without smart bins. Data is proposed to be communicated through a long-range, wide-area network (LoRaWAN). Another study proposed a system employing smart bins that measure the waste fill level to be collected by level 2 autonomous vehicles, where the communications were completed using a long-range wide area network (LoRaWAN), i.e., low power wide area (LPWA) instead of Wi-Fi and Ethernet (Karatagi et al., 2022 ). 2.5. Research Gap After more than a decade of research in this field, no fully functional smart waste collection system has been installed. Based on the reviewed literature to date, no study has utilized artificial intelligence and big data analytics to minimize the need for the typical hardware components of smart waste collection systems. Additionally, to the best of the authors' knowledge, the prediction of daily waste generation per bin is yet to be implemented. Moreover, driverless waste collection remains a contemporary research field that warrants significant attention, as intelligent transportation gradually emerges in smart cities. This is where a newly developed plan incorporating a fully automated system and an optimum routing strategy could have significant benefits. 2.6. Aim and Objectives This research presents a novel approach to smart waste collection by integrating multiple state-of-the-art technologies, including artificial intelligence, predictive analytics, route optimization, and intelligent transportation. This study aims to develop a dynamic framework for an autonomous, knowledge-based smart waste collection system, deploy the envisioned system in a computer simulation of a residential area over a one-month period, and assess its economic feasibility using a life-cycle costing (LCC) analysis. Historical records were used to determine the daily amount of waste generated per bin, acquired from onboard truck sensors, to predict future waste generation and accordingly select the bins that require collection. Next, collection routes were optimized for waste trucks and simulated the proposed system implementation using autonomous vehicles. Such enhancements to the waste collection system will significantly improve its economics, particularly in terms of capital investments and operational and maintenance costs, thereby fostering the field implementation of smart collection systems due to improved profitability and functionality. Mitigating health risks and environmental hazards associated with inadequate waste collection services and traffic congestion would add to the social benefits of such a system. The proposed approach will provide government waste management authorities and private waste collection companies with an alternative, sustainable method to deliver waste management services in smart cities. 3. Methodology This research developed a multidisciplinary framework for an autonomous knowledge-based smart waste collection system and simulated its implementation in an actual residential area to collect municipal solid waste. The following subsections present the system architecture, simulation data description, and methodological procedures for the various system components, including waste generation prediction, heuristic bin selection, route optimization, autonomous navigation, and LCC analysis. 3.1. System Architecture This study aims to develop a new autonomous smart waste collection system that achieves the most efficient operation, maximum economic benefits, and minimum environmental impacts. Figure 1 shows the framework of the proposed system. The primary data utilized by the system are historical records of collected waste per bin, including bin locations, collection routes, and daily quantities measured via truck onboard weighing sensors and GPS systems. In addition, the research utilizes various influential parameters, including 1) types of households, 2) demographic statistics, 3) seasonal calendars, and 4) other local data affecting waste generation patterns. As shown in Fig. 1 , the study used the obtained databases to train machine learning models to forecast the daily waste amounts at the bin level while incorporating deep reinforcement methods for supervised learning. Subsequently, a bin-selection algorithm selects bins to be serviced based on the predicted waste generation and maximum bin capacity. The model feeds designated bins to a route optimization module, which receives live traffic data simultaneously. The module transmits the optimized collection route to a GPS-enabled autonomous collection truck. Simultaneously, the system fine-tunes the models and software modules based on the actual quantity of generated waste measured through an onboard weighing system. 3.2. Simulation Description A computer simulation was conducted to evaluate the various components of the proposed system in an actual residential community located in Abu Dhabi, the second-most populous city in the United Arab Emirates (UAE), after Dubai. Figure 2 shows a map of the street layout and bin locations in the study area. The community comprises residential, institutional, and governmental buildings, with 375 single-detached dwellings and 12 institutional and commercial buildings, all connected by a street network with a total length of approximately 14 km. The Abu Dhabi Waste Management Center (Tadweer) provided the daily amount of recyclable waste per bin, i.e., paper, plastic, metal, and glass, for 396 bins with a volume of 240 liters each. Recyclable waste was selected as the stream for which collection routes are optimized due to hot weather conditions that prevent other types of waste, such as organics, from being retained in bins for more than a day. The dataset spanned the period from January 2019 to December 2020, yielding approximately 290,000 observations. Each instance included the day of collection, the latitude and longitude position of each bin, and the amount of waste in kilograms measured by onboard truck sensors. A unique number was assigned to each bin to facilitate the application of the bin selection algorithm. 3.3. Proposed Systems and Scenarios This study aims to compare existing traditional waste collection practices with sensor-based and knowledge-based smart waste collection systems. The study established three scenarios: 1) conventional waste collection (CWC), 2) sensor-based waste collection (SWC), and 3) knowledge-based waste collection (KWC). CWC represents the conventional waste collection system in which all bins are serviced daily through the same route. SWC, representing typical smart waste collection systems, employs weight sensors in the waste bins to obtain the amount of waste. Alternatively, the proposed knowledge-based system, i.e., KWC, utilizes historical records of daily generation patterns per bin, obtained from weight sensors onboard the collection trucks, to predict the produced quantities. The SWC and KWC systems collect only the bins that require service; hence, the collection route varies daily. It is expected that the SWC will be the best-performing system due to the acquired real-time measurements. However, it is an expensive system consisting of various costly components and intricate wireless connections that require multiple modifications to the city infrastructure and increased maintenance costs. 3.4. Waste Generation Prediction Figure 3 presents the framework of the various processes implemented to forecast the daily waste generated per bin. Data preprocessing was performed by cleaning and transforming the raw data compiled from the waste collection company. This involved removing duplicate records, handling missing or inconsistent entries, and filtering out bins with incomplete location data. Extreme outliers, primarily caused by sensor malfunctions, were identified using the interquartile range method. Moreover, categorical variables such as day of week and season were encoded into numerical features to capture temporal variation. The dataset was split into 67% and 33% subsets for training and testing, respectively. These fractions correspond to model training from January 4, 2019, until March 31, 2020, with testing starting the following day. Feature engineering, including feature generation and selection, was implemented to eliminate redundant or irrelevant features from the dataset, thereby enhancing accuracy and facilitating a better understanding of the underlying relationship between dependent and independent variables. The study trained two machine learning algorithms, a generalized linear model (GLM) and gradient-boosted trees (XGBoost), on historical daily waste generation data per bin. A GLM is an extension of conventional linear models, combining both linear and logistic regression, in which the data are fitted by following a maximum likelihood approach (Zheng & Agresti, 2000 ). It offers multiple regression advantages within a flexible framework; hence, the model-fitting computation is rapid and parallel. Moreover, a GLM is adequate for models with missing values and a limited number of nonzero parameter predictors. On the other hand, XGBoost is a set of classification or regression decision trees that employ a flexible nonlinear regression procedure, providing a forward-learning ensemble method to obtain predictive results through incrementally improved estimations. XGBoost effectively enhances accuracy due to its sensitivity to the computational range of data and features; thus, it does not require data normalization (Sun Yin & Vatrapu, 2017 ). These two models were selected to represent complementary approaches, where the GLM serves as a transparent baseline model that captures linear dependencies, and XGBoost is an advanced ensemble method capable of modeling nonlinear relationships. This combination enabled benchmarking a simple interpretable model against a state-of-the-art machine learning algorithm. This study utilized RapidMiner Studio software to develop the selected models. Each model was trained and optimized by running multiple iterations to determine the optimal parameters. Hyperparameter optimization was carried out through grid search or parameter sweep, the most applied method for searching through a predefined subset of the learning algorithm hyperparameter space. The study specified these parameters using minimum and maximum values, i.e., lower and upper bounds, along with the number of incremental steps with different scales, e.g., linear, logarithmic, and quadratic scales. In a grid search, trials were established by aggregating every conceivable value combination. This study employed a 3-fold cross-validation (CV) and grid search optimization approach to optimize three parameters of the XGBoost model: the number of trees, maximal depth, and learning rate, with ranges of 30–150, 2–7, and 0.001-0.1, respectively. The optimal number of trees, maximum depth, and learning rate were found to be 150, 7, and 0.1, respectively. Afterward, a CV method was deployed to test the models. Evaluation metrics were applied to assess the performance of the developed waste prediction models, specifically the root mean squared error (RMSE), squared error (SE), absolute error (AE), relative error (RE), and R-squared (R 2 ). 3.5. Bin-selection Algorithm The three examined scenarios were modeled using the MATLAB programming platform, following the diagram shown in Fig. 4 . The bin-selection algorithm utilized the predicted amount of waste in each bin obtained from the best-performing AI model to select the bins to be collected. A heuristic system determined whether the predicted waste volume exceeded the effective bin capacity. If so, it assigned the bin for collection. Otherwise, the system skipped the bin along the collection route and repeated the capacity check for the same bin the next day, and so on. For instance, the CWC scenario represented the traditional system that collects all the bins every day. On the other hand, the bin-selection algorithm in the SWC and KWC scenarios selected only the bins that reached 90% of their maximum capacity, i.e., considering a 10% safety margin. On any given day, the amount of waste in bins that were not collected accumulates. 3.6. Route Optimization The system used the selected bins acquired from the bin-selection algorithm to optimize the route for waste collection trucks. Route optimization was implemented using street locations and the position of the selected bins in terms of latitude and longitude coordinates. The Network Analyst extension in ArcMap software was utilized to optimize routes based on the Dijkstra algorithm, which determines the shortest path in terms of total travel distance. For instance, the system considered all possible routes between bins and selected the one with the minimum length. The system performed route optimization only once to collect all bins in the CWC scenario, whereas the SWC and KWC scenarios applied route optimization daily, as the selected bins changed daily in response to variations in waste generation. 3.7. Autonomous Navigation This research carried out traffic modeling and simulation using PTV Vissim software. The analysis comprises three main steps: 1) establish layout and geometry considerations for the study area, 2) compile traffic data for the major and minor roads on weekdays and weekends, and 3) conduct ten simulation runs over a month to produce a reliable estimation due to the stochastic nature of the simulation. The study randomly simulated the time spent at each bin following a normal distribution with a mean of 60 seconds and a standard deviation of 30 seconds. The waste collection truck route in the CWC, SWC, and KWC scenarios was calibrated and validated using human driving behavior to reflect the actual conditions of the network categories, as well as predicted autonomous driving behavior to ensure realistic forecasting of their impacts. Two driving behavior characteristics were collected using digital wave radar detectors, standstill distance and headway (CC0 and CC1 in Vissim, respectively). Three driving behavior models in Vissim were considered: car-following, lane-changing, and lateral behavior within the lane. The parameters of all three models were modified and validated using real-world trajectory data collected from a public test track. Data collection and calibration of CAV behavior in traffic were performed as part of the CoExist project (CoExist, 2019 ). The parameters were modified to simulate the ability of CAV to maintain a shorter headway, make instant reaction decisions, and achieve harmonized average speeds. In addition, the ability to communicate with other vehicles was simulated by enabling the waste collection truck to adjust its speed and gap acceptance preferences based on the types of surrounding vehicles. A sensitivity analysis was conducted to investigate the impact of varying traffic volume and operating speed on traffic delay, throughput, and carbon footprint. The selected factors can vary due to various reasons, such as traffic congestion at peak hours, extreme climatic conditions, governmental restrictions, and road accidents. A traffic volume of 70,000 vehicles and an operating speed of 45 km/h were utilized, with variations of 20,000 vehicles and ± 5 km/h, respectively. 3.8. Life Cycle Costing An LCC was conducted to assess the economic performance of the examined waste collection systems. The analysis aimed to quantify the net present cost (NPC) of each system over a 15-year study period at a 5% discount rate. The capital expenditure (CAPEX) covered the procurement, installation, and periodic replacement of physical assets over their respective lifetimes, whereas the operational expenditure (OPEX) included fuel, labor, and maintenance costs derived from the simulated travel distances and service frequencies. Table 3 breaks down the CAPEX and OPEX input parameters. The unit costs were obtained from international vendors at market prices to ensure realistic estimates. Table 3 Cost streams of the examined scenarios. Parameter CWC SWC KWC Bin ( $ ) 80 80 80 Lifespan of the bin (year) 5 5 5 Bin smart sensor ( $ ) - 70 - Processing and connectivity module ( $ ) - 20 - Lifespan of bin instrumentation (year) - 5 - Maintenance of equipped bin (% CAPEX) - 20 - Waste collection vehicle ( $ ) 60,000 60,000 60,000 Instrumentation of autonomous vehicle ( $ ) 10,000 10,000 10,000 Onboard vehicle weighing system ( $ ) - - 3,000 Lifespan of onboard vehicle weighing system (year) - - 5 Maintenance of equipped vehicle (% CAPEX) - - 20 Hourly rate of heavy vehicle driver ( $ /h) 20 20 20 4. Results and Discussion This section discusses the outcomes of the simulation analyses for data processing, waste generation forecasting, bin selection, route optimization, and autonomous operation phases, followed by a financial assessment. This discussion aims to compare the proposed knowledge-based autonomous smart collection system with typical sensor-based smart and traditional collection systems. This section investigates three scenarios: conventional waste collection, sensor-based waste collection, and knowledge-based waste collection. 4.1. Data Processing Figure 5 shows the daily waste generation pattern in the study area during the analysis period. Waste generation fluctuated significantly daily, with higher and more variable quantities produced on average in 2019 compared to 2020. The daily generation in 2019 ranged from − 19.6% to 31.2% of the mean value of 4.7 tons, whereas it ranged between − 21.8% and 17.3% of the average value of 3.2 tons in 2020. The maximum generation in 2019 occurred in May and September, corresponding to the holy month of Ramadan and the beginning of the academic year, respectively. In contrast, the lowest amounts of waste were generated in January and August, during the study breaks. The waste generation in 2020 did not follow a pattern similar to that in 2019, as the lowest quantities were generated in May, whereas the highest waste production occurred in January, July, and August. 4.2. Waste Generation Prediction Predicting waste generation is one of the key differences between the proposed knowledge-based system, represented by the KWC scenario, and the sensor-based SWC scenario. The waste generation predictions, based on historical data acquired from onboard truck sensors, aim to replace the real-time data measured by bin-level or weight sensors. Table 4 lists the prediction results of the examined machine learning models to be used in the KWC scenario. The results indicated that both models efficiently predicted actual waste generation with R 2 values of 0.94 and 0.92 for XGBoost and the GLM, respectively. However, optimized XGBoost demonstrated a higher degree of accuracy than the GLM, showing a considerable enhancement in terms of the AE and RE by approximately 32% and 21%, respectively. Moreover, slight improvements of approximately 0.4% and 1.8% in the RMSE and SE, respectively, were observed. Table 4: Comparative assessment of the prediction models in terms of various evaluation metrics. Model RMSE AE RE SE R 2 Generalized Linear Model (GLM) 0.470 0.330 5.24% 0.224 0.92 Gradient Boosted Trees (XGBoost) 0.468 0.224 4.16% 0.220 0.94 Root Mean Squared Error (RMSE), Absolute Error (AE), Relative Error (RE), Squared Error (SE), Coefficient of Determination (R 2 ) 4.3. Bin-Selection Algorithm The heuristic bin-selection algorithm was applied in the SWC and KWC scenarios using measured and predicted waste data, respectively. Figure 7 shows the daily variation in the number of serviced bins and the amount of waste collected in all scenarios. The number of bins and the amount of waste in the SWC and KWC scenarios varied significantly, with peak values occurring every 4–5 days. In contrast, the CWC scenario exhibited minimal deviations in managed waste quantities, with a constant number of serviced bins. Furthermore, both the number of bins and the amount of waste followed the same pattern for the SWC and KWC scenarios. Table 5 summarizes these variables in addition to the number of collection days. The CWC scenario had a fixed number of bins, approximately 396, collected every day, totaling 11,880 bins serviced in a month. In contrast, the SWC and KWC systems provided collection on only 25 and 27 days, respectively, as the daily generated waste did not exceed the bin capacity on certain days. Consequently, the total number of collected bins decreased significantly by 89% for both scenarios; the KWC scenario had only two fewer bins than the SWC scenario, despite the difference in minimum and maximum values. The SWC and KWC systems did not require servicing any bins in the first three days; hence, the waste accumulated until it reached full capacity. Moreover, despite the significant reduction in the number of bins selected in the SWC and KWC scenarios compared to the conventional system, the amount of collected waste was reduced by only approximately 9.97% and 10.61%, respectively. This highlights the potential improvements to conventional waste collection that can be achieved by replacing it with smart systems, which offer a similar level of service with significantly greater benefits. Table 5 Statistical bin-selection algorithm measures for the examined scenarios. Variable CWC SWC KWC Number of Bins Serviced Daily Minimum 396 11 17 Maximum 396 147 116 Total 11,880 1,288 1,286 Daily Amount of Waste (ton) Minimum 2.48 0.59 0.92 Maximum 3.10 8.50 6.65 Total 83.03 74.76 74.22 Number of Collection Days Total 30 25 27 CWC: Conventional Waste Collection, SWC: Sensors-based Waste Collection, KWC: Knowledge-based Waste Collection 4.4. Route Optimization The waste collection route was optimized for all scenarios, once for the CWC scenario and daily for the SWC and KWC scenarios. Figure 8 displays the daily variations in the optimum route length for the SWC and KWC scenarios, whereas the CWC scenario had a constant daily travel distance because all bins were collected daily. The bin-selection algorithm showed that the system collected no waste on certain days in the SWC and KWC scenarios; consequently, the travel distance was zero. This resulted in an approximately 7 km decrease in the total SWC travel distance compared to the KWC scenario, as the latter required collection on extra days, although it had a slightly lower number of serviced bins. Figure 9 shows the optimized CWC waste collection route, as well as the low- and high-service SWC and KWC days, along with several statistical measures of the travel distance for all scenarios. The travel distance of CWC was found to be 10.9 km, whereas it ranged between 1.5 and 8.4 km and 2.7 and 7.5 km for the SWC and KWC scenarios, respectively. In terms of the total travel distance, the SWC and KWC scenarios significantly outperformed the CWC scenario by 63.1% and 60.9%, respectively. 4.5. Autonomous Navigation This study simulated autonomous navigation for all scenarios to investigate its individual effect on conventional and smart waste collection systems. Table 6 illustrates the application of CAVs against human-driven vehicles in all assessed scenarios. The CWC scenario had the highest carbon footprint, whereas the SWC and KWC scenarios demonstrated significantly better performance, minimizing total emissions by 49% and 48% for human-driven vehicles and 49% and 50% for CAVs, respectively, due to their considerably lower total travel distance and number of collection days. In addition, the total emissions were decreased by CAVs in all scenarios, with the KWC scenario achieving the highest reduction of 29%. Similarly, the CWC scenario resulted in the longest travel time, while the SWC and KWC scenarios significantly reduced the total travel time by approximately 87% and 85%, respectively, for both regular and autonomous vehicles. In contrast, regular and autonomous vehicles demonstrated a similar performance, as there was no traffic congestion. Moreover, the SWC and KWC scenarios experienced no or low traffic delays, as several days did not require collections, and queued vehicles did not obstruct the waste collection trucks due to relatively shorter routes compared to the CWC scenario, which had the longest traffic delays. However, the maximum delay of the CWC scenario was 3.24 seconds using CAVs, outperforming the SWC scenario by 67%. The implementation of CAVs significantly improved the average traffic delay by 83%, 68%, and 73% for the CWC, SWC, and KWC scenarios, respectively. Furthermore, the CWC scenario resulted in the longest queue length, whereas the SWC and KWC scenarios achieved reductions of 20% and 11%, respectively, using human-driven collection vehicles. These reductions approximately doubled when using CAVs, with values of approximately 42% and 21%, respectively. Additionally, autonomous collection trucks enhanced the average queue length by 46%, 61%, and 52% in the CWC, SWC, and KWC scenarios, respectively. Table 6: Traffic simulation results for regular and autonomous vehicles. Variable Vehicle Type CWC SWC KWC Travel Time (h) Regular Minimum 6.58 0.48 0.51 Maximum 7.04 1.99 1.68 Total 173.91 22.94 26.32 Autonomous Minimum 6.53 0.46 0.52 Maximum 6.84 1.91 1.58 Total 172.75 22.39 25.88 Queue Length (m) Regular Minimum 26.43 28.14 24.11 Maximum 42.85 40.60 43.16 Average 34.87 28.06 31.07 Autonomous Minimum 11.32 11.18 10.35 Maximum 23.57 18.28 18.95 Average 18.84 10.99 14.95 Traffic Delay (s) Regular Minimum 17.42 2.87 2.79 Maximum 19.93 11.33 11.38 Average 18.63 5.40 6.82 Autonomous Minimum 2.97 0.84 1.68 Maximum 3.24 5.41 2.35 Average 3.12 1.73 1.86 Carbon Footprint (CO 2 -eq) Regular Minimum 0.94 0.44 0.33 Maximum 1.31 0.92 0.80 Total 28.22 14.29 14.69 Autonomous Minimum 0.68 0.34 0.23 Maximum 1.01 0.68 0.62 Total 21.16 10.69 10.50 CWC: Conventional Waste Collection, SWC: Sensors-based Waste Collection, KWC: Knowledge-based Waste Collection Figure 10 illustrates the impact of varying the traffic volume and operating speed on traffic delay, throughput, and carbon footprint in all scenarios using both regular and autonomous waste collection vehicles. The model reveals significant changes when traffic volume is increased, particularly in terms of traffic delay and carbon footprint, compared to when it is reduced. For instance, the percentage change in the traffic delay and carbon footprint when increasing the traffic volume for all scenarios ranged from -260% to -100% and from -200% to -70%, respectively, whereas the maximum change realized by decreasing the traffic volume reached approximately 50% and 30%, respectively. Overall, the impact of CAVs was more significant at higher traffic volumes and lower operating speeds, as the benefits of CAVs cannot be fully realized in congested traffic. Figure 11 shows a heatmap of the traffic volume, in vehicle per hour (vph), for the most utilized streets in the study area for the CWC scenario, as well as the low- and high-service days of the SWC and KWC scenarios. The CWC scenario had the most congested roads, as all bins were collected, followed by the SWC and KWC scenarios, which required fewer bins to be collected. 4.6. Economic Feasibility Figure 12 presents the annual NPC for the scenarios examined in the study area over a 15-year evaluation period. The investment costs included capital expenditures for waste collection vehicles, bins, and their related instrumentation. The periodic cost spikes every five years correspond to the scheduled replacement of bins and sensors in accordance with their expected service life. The CWC scenario had an NPC of approximately $ 580k, mainly due to high fuel consumption and labor intensity from daily fixed-route operations. The SWC system incurred the highest NPC, at approximately $ 609k; although optimized routing reduced fuel usage, the savings were insufficient to offset the significant expenses of sensors, connectivity modules, and their recurring maintenance, resulting in a total cost 5% higher than that of CWC. On the other hand, despite using a more costly sensor, the KWC system achieved a significantly lower NPC of $ 196k, representing a 65% reduction relative to CWC and SWC. This substantial reduction is primarily attributed to the elimination of hundreds of bin-mounted sensors and their associated installation and maintenance costs. By centralizing data collection through a single onboard weighing system and leveraging predictive, cloud-based analytics, KWC minimized hardware dependence while maintaining high operational efficiency. When CAV configuration was introduced, the cost of vehicle electrification was incorporated, while fuel and labor costs were reduced due to automation and shorter collection times. Despite these operational savings, CAV scenarios exhibited slightly higher NPCs compared to their conventional counterparts: $ 586k, $ 627k, and $ 223k for CWC-CAV, SWC-CAV, and KWC-CAV, respectively. The increase is mainly attributed to the added costs of instrumentation for autonomous vehicles. These findings underscore the significant economic and operational benefits of data-driven, cloud-based architectures over hardware-intensive systems. 4.7. Overall Assessment Figure 13 compares the outcomes of the bin-selection algorithm, route optimization, and autonomous operation analyses for the smart waste collection scenarios as a percentage reduction from the baseline scenario (CWC). Similar results were obtained using both regular and autonomous vehicles in the bin-selection algorithm, and route optimization analyses were assessed in terms of autonomous operation through metrics such as travel time, queue length, traffic delay, carbon footprint, and economic feasibility. The SWC and KWC systems reduced the total number of collected bins by approximately 89%, as the bin-selection algorithm selected only full bins. This led to avoiding collection on five and three days in the SWC and KWC scenarios, respectively, as no bins required servicing; thus, the amount of collected waste was approximately 10% lower in both scenarios. In addition, the total travel distance decreased by 63% and 61% in the SWC and KWC scenarios, respectively, as the daily number of selected bins was consistently less than that in the CWC scenario. Moreover, the total travel time of both regular and autonomous vehicles decreased by approximately 87% and 85% for the SWC and KWC scenarios, respectively. Using regular vehicles in the SWC and KWC scenarios increased the carbon footprint by 49% and 48%, respectively, compared to the conventional scenario, whereas CWC automation led to a 25% improvement. However, the latter contributed to an 83% reduction in traffic delay and a 46% reduction in queue length, respectively. From an economic perspective, KWC achieved substantially lower NPC by 63%, whereas SWC incurred 7% higher costs due to the additional hardware and maintenance expenses. In general, the automation slightly increased the NPC due to the added vehicle instrumentation costs, which were not fully offset by reduced operation expenses. Overall, the best performance in terms of traffic delay, queue length, and carbon footprint was realized by utilizing autonomous vehicles in the SWC and KWC scenarios. 4.8. Potential Applications and Challenges Beyond the waste management sector, this framework can be adopted in a broad range of applications, as supply forecasting, resource selection, and route optimization are not unique to waste logistics. Similar architectures could be tailored to other industries where demand is variable and spatially distributed, such as parcel delivery, food distribution, or recycling of electronic waste. Even outside urban services, the same principles could be applied to industrial supply chains; for instance, optimizing the collection of agricultural produce, medical supplies, or construction debris. Embedding predictive analytics and autonomous routing into such domains highlights the versatility of the framework and its potential contribution to smart city systems and industrial sustainability goals. Nevertheless, from a practical implementation perspective, several challenges may arise, particularly in scaling the system from simulation to large-scale deployment. For example, the maintenance of onboard sensors, the integration of autonomous vehicles into existing traffic regulations, and the interoperability between municipal platforms are all issues that operators must address before widespread adoption. Real-world pilot projects would be required to evaluate robustness under operational uncertainties, such as irregular waste generation, extreme weather conditions, or hardware malfunctions. Nevertheless, the proposed system reduces reliance on bin sensors, which lowers infrastructure costs and improves scalability compared to traditional hardware‐intensive approaches. 5. Conclusion A novel autonomous cloud-based smart waste collection system has been proposed to replace conventional systems, reducing capital investments, operating costs, traffic congestion, and air pollution emissions. The system integrates various components: 1) AI models predicting the daily waste generation for each bin, 2) a bin-selection algorithm prioritizing bins requiring collection, 3) optimization techniques identifying the best collection route, and 4) autonomous navigation providing driverless waste collection. The proposed framework was applied through computer simulations of a residential area in Abu Dhabi, UAE, which contained 396 bins. For this area, 2-year historical records of daily generated recyclable waste per bin were obtained. Three scenarios were compared: CWC, SWC, and KWC. CWC represented the conventional system, in which all bins were collected daily through the same route. SWC represented a typical sensor-based smart system, and KWC demonstrated the proposed system. The GLM and XGBoost models were tested to predict daily waste quantities, where the latter outperformed the former with an absolute error of 0.22. The KWC and SWC scenarios utilized waste quantities forecasted by the XGBoost model and the actual quantities generated, respectively, to select the bins requiring collection using a bin-selection algorithm. Both the SWC and KWC achieved significant reductions in the total number of serviced bins, selecting 89% fewer bins than the CWC. Next, the collection routes for each scenario were optimized using GIS methods to determine the shortest travel path. The total CWC travel distance improved by 63% and 61% in the SWC and KWC scenarios, respectively. The operation of autonomous vehicles was analyzed and compared to regular driving in all examined scenarios. The total delay improved significantly by 91% and 90% in the automated SWC and KWC scenarios, respectively, whereas the CWC scenario had the longest traffic delay and travel time. Moreover, an LCC analysis was conducted to assess the economic performance of the systems over a 15-year period. The results revealed that SWC incurred the highest costs due to extensive sensor installation and maintenance, while KWC achieved a 65% cost reduction relative to SWC by replacing hardware-intensive components with a cloud-based system. Overall, the technical performance of the KWC and SWC systems was close; nevertheless, the KWC system is distinctly advantageous over the SWC system because it eliminates the capital and operating costs of the bin sensory data acquisition and communication system, offering a more cost-effective, eco-friendly, and less complex smart system to establish and maintain. Declarations Funding This research received no specific grant from any funding agency in the public, private, or not-for-profit sectors. Author Contribution M.A: conceptualization, formal analysis, methodology, data curation, supervision, validation, and writing—review and editing, and M.H: formal analysis, methodology, software, writing—original draft. Data Availability The data supporting the findings of this study were provided by Abu Dhabi Waste Management Center (Tadweer), but restrictions apply to the availability of the data. The daily waste generation records, used under license from Tadweer for the purpose of this study, are not publicly available. References Abbasi, M., Rastgoo, M. N. & Nakisa, B. Monthly and seasonal modeling of municipal waste generation using radial basis function neural network. Environ. Progress Sustainable Energy . 38 (3). https://doi.org/10.1002/ep.13033 (2019). Abdallah, M., Adghim, M., Maraqa, M. & Aldahab, E. Simulation and optimization of dynamic waste collection routes. Waste Manage. Res. 37 (8), 793–802. https://doi.org/10.1177/0734242X19833152 (2019). Adusei, K. K., Ng, K. T. W., Mahmud, T. S., Karimi, N. & Lakhan, C. Exploring the use of astronomical seasons in municipal solid waste disposal rates modeling. Sustainable Cities and Society , 86 . (2022). https://doi.org/10.1016/j.scs.2022.104115 Alfeo, A. L. et al. Urban swarms: A new approach for autonomous waste management. Proceedings - IEEE International Conference on Robotics and Automation , 2019-May , 4233–4240. (2019). https://doi.org/10.1109/ICRA.2019.8794020 Alsobky, A., Ahmed, M., Agroudy, A., Araby, E. & S., &, K A smart framework for municipal solid waste collection management: A case study in Greater Cairo Region. Ain Shams Eng. J. 14 (6). https://doi.org/10.1016/j.asej.2023.102183 (2023). Amal, L., Son, L. H. & Chabchoub, H. SGA: spatial GIS-based genetic algorithm for route optimization of municipal solid waste collection. Environ. Sci. Pollut. Res. 25 (27), 27569–27582. https://doi.org/10.1007/s11356-018-2826-0 (2018). Araiza-Aguilar, J. A., Rojas-Valencia, M. N. & Aguilar-Vera, R. A. Forecast generation model of municipal solid waste using multiple linear regression. Global J. Environ. Sci. Manage. 6 (1), 1–14. https://doi.org/10.22034/gjesm.2020.01.01 (2020). Belhiah, M., El Aboudi, M. & Ziti, S. Optimising unplanned waste collection: An IoT-enabled system for smart cities, a case study in Tangier, Morocco. IET Smart Cities . 6 (1), 27–40. https://doi.org/10.1049/smc2.12069 (2024). Brouwer, Y., Barbosa-Póvoa, A. P., Antunes, A. P., Rodrigues, P. & Ramos, T. Comparison of different waste bin monitoring approaches: An exploratory study. Waste Manage. Res. 41 (10), 1570–1583. https://doi.org/10.1177/0734242X231160691 (2023). Catania, V. & Ventura, D. An approch for monitoring and smart planning of urban solid waste management using smart-M3 platform. Conference of Open Innovation Association, FRUCT , 24–31. (2014). https://doi.org/10.1109/FRUCT.2014.6872422 CoExist. PTV Vissim Tool . (2019). Fan, L. et al. Introducing an evolutionary-decomposition model for prediction of municipal solid waste flow: application of intrinsic time-scale decomposition algorithm. Eng. Appl. Comput. Fluid Mech. 15 (1), 1159–1175. https://doi.org/10.1080/19942060.2021.1945496 (2021). Ghanbari, F., Kamalan, H. & Sarraf, A. An evolutionary machine learning approach for municipal solid waste generation estimation utilizing socioeconomic components. Arab. J. Geosci. https://doi.org/10.1007/s12517-020-06348-w/Published (2021). Greco, S. A. M., Rossit, D. G., Frutos, M. & Cavallin, A. Optimization of waste collection through the sequencing of micro-routes and transfer station convenience analysis: an Argentinian case study. Waste Management & Research . (2023). https://gobiernoabierto.bahia.gob.ar/ambientales/residuos-domiciliarios/ Guha, B., Momtaz, Z., Kafy, A., Al & Rahaman, Z. A. Estimating solid waste generation and suitability analysis of landfill sites using regression, geospatial, and remote sensing techniques in Rangpur, Bangladesh. Environ. Monit. Assess. 195 (1). https://doi.org/10.1007/s10661-022-10695-4 (2023). Gutierrez, J. M., Jensen, M., Henius, M. & Riaz, T. Smart Waste Collection System Based on Location Intelligence. Procedia Comput. Sci. 61 , 120–127. https://doi.org/10.1016/j.procs.2015.09.170 (2015). Hoang, M. G., Fujiwara, T., Pham Phu, S. T. & Nguyen Thi, K. T. Predicting waste generation using Bayesian model averaging. Global J. Environ. Sci. Manage. 3 (4), 385–402. https://doi.org/10.22034/gjesm.2017.03.04.005 (2017). Hu, S., An, L. & Shen, L. A multi-objective modeling and optimization approach to municipal solid waste collection for classified treatment in China towards sustainable development. Sustainable Cities and Society , 98 . (2023). https://doi.org/10.1016/j.scs.2023.104846 Hussain, D. I., Elomri, D. A., Kerbache, D. L., Omri, D. A. & El Smart city solutions: Comparative analysis of waste management models in IoT-enabled environments using multiagent simulation. Sustainable Cities and Society , 103 . (2024). https://doi.org/10.1016/j.scs.2024.105247 Jassim, M. S., Coskuner, G. & Zontul, M. Comparative performance analysis of support vector regression and artificial neural network for prediction of municipal solid waste generation. Waste Manage. Res. 40 (2), 195–204. https://doi.org/10.1177/0734242X211008526 (2022). Jorge, D., Pais Antunes, A., Rodrigues Pereira Ramos, T. & Barbosa-Póvoa, A. P. A hybrid metaheuristic for smart waste collection problems with workload concerns. Computers and Operations Research , 137 . (2022). https://doi.org/10.1016/j.cor.2021.105518 Justo, V. B., Gupta, A., Umland, T. F. & Göhlich, D. Minimum Energy Utilization Strategy for Fleet of Autonomous Robots in Urban Waste Management. Robotics 12 (6). https://doi.org/10.3390/robotics12060159 (2023). Juwariyah, T., Krisnawati, L. & Sulasminingsih, S. Design of IoT-Based Smart Bins Integrated Monitoring System Using Blynk. IOP Conference Series: Materials Science and Engineering , 1125 (1), 012078. (2021). https://doi.org/10.1088/1757-899x/1125/1/012078 Karatagi, A., Babu Peter, R., Shubham, R., Biradar, R. C. & Devanagavi, G. D. Integrating Electric Vehicle (EV) in Waste Management for Garbage Collection using LoRa Technology. Proceedings – 2022 6th International Conference on Intelligent Computing and Control Systems, ICICCS 2022 , 121–128. (2022). https://doi.org/10.1109/ICICCS53718.2022.9788416 Kaya, S. A hybrid firefly and particle swarm optimization algorithm with local search for the problem of municipal solid waste collection: a real-life example. Neural Comput. Appl. 35 (9), 7107–7124. https://doi.org/10.1007/s00521-022-08173-6 (2023). Kontokosta, C. E., Hong, B., Johnson, N. E. & Starobin, D. Using machine learning and small area estimation to predict building-level municipal solid waste generation in cities. Comput. Environ. Urban Syst. 70 , 151–162. https://doi.org/10.1016/j.compenvurbsys.2018.03.004 (2018). Latif, S. D., Hazrin, N. A. B., Younes, M. K., Ahmed, A. N. & Elshafie, A. Evaluating different machine learning models for predicting municipal solid waste generation: a case study of Malaysia. Environ. Dev. Sustain. https://doi.org/10.1007/s10668-023-03882-x (2023). Liang, G. et al. Predicting municipal solid waste using a coupled artificial neural network with archimedes optimisation algorithm and socioeconomic components. Journal of Cleaner Production , 315 . (2021). https://doi.org/10.1016/j.jclepro.2021.128039 Lu, W., Huo, W., Gulina, H. & Pan, C. Development of machine learning multi-city model for municipal solid waste generation prediction. Front. Environ. Sci. Eng. 16 (9). https://doi.org/10.1007/s11783-022-1551-6 (2022). Luo, K., Zhao, W. & Zhang, R. A multi-day waste collection and transportation problem with selective collection and split delivery. Appl. Math. Model. 126 , 753–771. https://doi.org/10.1016/j.apm.2023.11.009 (2024). Mahéo, A., Rossit, D. G. & Kilby, P. Solving the integrated bin allocation and collection routing problem for municipal solid waste: a Benders decomposition approach. Ann. Oper. Res. 322 (1), 441–465. https://doi.org/10.1007/s10479-022-04918-7 (2023). Mamun, M. A., Al, Hannan, M. A., Hussain, A. & Basri, H. Theoretical model and implementation of a real time intelligent bin status monitoring system using rule based decision algorithms. Expert Syst. Appl. 48 , 76–88. https://doi.org/10.1016/j.eswa.2015.11.025 (2016). Martikkala, A., Mayanti, B., Helo, P., Lobov, A. & Ituarte, I. F. Smart textile waste collection system – Dynamic route optimization with IoT. Journal of Environmental Management , 335 . (2023). https://doi.org/10.1016/j.jenvman.2023.117548 Mohammadi, M. et al. A dynamic approach for the multi-compartment vehicle routing problem in waste management. Renewable and Sustainable Energy Reviews , 184 . (2023). https://doi.org/10.1016/j.rser.2023.113526 Mudannayake, O., Rathnayake, D., Herath, J. D., Fernando, D. K. & Fernando, M. Exploring Machine Learning and Deep Learning Approaches for Multi-Step Forecasting in Municipal Solid Waste Generation. IEEE Access. 10 , 122570–122585. https://doi.org/10.1109/ACCESS.2022.3221941 (2022). Namoun, A. et al. An Ensemble Learning Based Classification Approach for the Prediction of Household Solid Waste Generation. Sensors 22 (9). https://doi.org/10.3390/s22093506 (2022). Noufal, M., Yuanyuan, L., Maalla, Z. & Adipah, S. Determinants of household solid waste generation and composition in Homs city, Syria. Journal of Environmental and Public Health , 2020 . (2020). https://doi.org/10.1155/2020/7460356 Rambandara, R. D. S. S., Prabodanie, R. A. R., Karunarathne, E. A. C. P. & Rajapaksha, R. D. D. Improving the Efficiency of Urban Waste Collection Using Optimization: a Case Study. Process. Integr. Optim. Sustain. 6 (3), 809–818. https://doi.org/10.1007/s41660-022-00232-8 (2022). Ramos, T. R. P., de Morais, C. S. & Barbosa-Póvoa, A. P. The smart waste collection routing problem: Alternative operational management approaches. Expert Syst. Appl. 103 , 146–158. https://doi.org/10.1016/j.eswa.2018.03.001 (2018). SAE. Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles . (2018). Sidhu, N., Pons-Buttazzo, A., Muñoz, A. & Terroso-Saenz, F. A collaborative application for assisting the management of household plastic waste through smart bins: A case of study in the Philippines. Sensors 21 (13). https://doi.org/10.3390/s21134534 (2021). Song, J. et al. Simulated annealing based hybrid forecast for improving daily municipal solid waste generation prediction. Scientific World Journal , 2014 . (2014). https://doi.org/10.1155/2014/834357 Srivastava, A. & Jha, P. K. A multi-model forecasting approach for solid waste generation by integrating demographic and socioeconomic factors: a case study of Prayagraj, India. Environ. Monit. Assess. 195 (6). https://doi.org/10.1007/s10661-023-11338-y (2023). Sun Yin, H. & Vatrapu, R. A first estimation of the proportion of cybercriminal entities in the bitcoin ecosystem using supervised machine learning. Proceedings – 2017 IEEE International Conference on Big Data, Big Data 2017 , 2018-Janua , 3690–3699. (2017). https://doi.org/10.1109/BigData.2017.8258365 Zhang, S., Zhang, J., Zhao, Z. & Xin, C. Robust Optimization of Municipal Solid Waste Collection and Transportation with Uncertain Waste Output: A Case Study. J. Syst. Sci. Syst. Eng. 31 (2), 204–225. https://doi.org/10.1007/s11518-021-5510-8 (2022). Zheng, B. & Agresti, A. Summarizing the predictive power of a generalized linear model . December 1998 , 1771–1781. (2000). Additional Declarations No competing interests reported. Supplementary Files SupplementaryFiles.docx Cite Share Download PDF Status: Published Journal Publication published 26 Apr, 2026 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Revision requested 04 Dec, 2025 Reviews received at journal 03 Dec, 2025 Reviews received at journal 03 Dec, 2025 Reviews received at journal 01 Dec, 2025 Reviewers agreed at journal 27 Nov, 2025 Reviewers agreed at journal 24 Nov, 2025 Reviewers agreed at journal 23 Nov, 2025 Reviewers agreed at journal 22 Nov, 2025 Reviewers agreed at journal 06 Nov, 2025 Reviewers agreed at journal 06 Nov, 2025 Reviewers invited by journal 06 Nov, 2025 Editor invited by journal 06 Nov, 2025 Editor assigned by journal 04 Nov, 2025 Submission checks completed at journal 04 Nov, 2025 First submitted to journal 02 Nov, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8012685","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":545719636,"identity":"8ebdee58-f4b3-4c13-9cf9-a7c33f94fb3a","order_by":0,"name":"Mohamed Abdallah","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABCElEQVRIiWNgGAWjYFACHgYJhgILBgP2BjYgj43BgDgtBkDEc4BkLRIJbGA+QS267WcP3vhhICFvLvn82YOfe/jkzRmYH35gqLHBqcXsTF6yZY+BhOHO2Tnmhj3P2Ax3NrAZSzAcS8Ot5UCOmQSPgQTjhts5bBJA/zBuOMBgBvTTYdxazr8xk/xjIGG/4ebxZ5J/DrDZbzjA/o2B4d9/3Fpu5JhJA21J3HCDAcg4wJa44QCPGQNj2wE8Wt4lW8sYSCRvOAPUK3OALXnDYZ5iicS+ZDwOyz14802Fje2G40CHvTlwDMho3/jhwzc7nFrQwTEGBmYglUC0BgaGGhLUjoJRMApGwUgBAK99VDe3Wk69AAAAAElFTkSuQmCC","orcid":"","institution":"University of Sharjah","correspondingAuthor":true,"prefix":"","firstName":"Mohamed","middleName":"","lastName":"Abdallah","suffix":""},{"id":545719637,"identity":"11e0bd81-1270-42bd-a0d7-b88a874bd219","order_by":1,"name":"Mariam Hosny","email":"","orcid":"","institution":"University of Sharjah","correspondingAuthor":false,"prefix":"","firstName":"Mariam","middleName":"","lastName":"Hosny","suffix":""}],"badges":[],"createdAt":"2025-11-02 18:23:21","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8012685/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8012685/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-026-48792-w","type":"published","date":"2026-04-26T16:00:16+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":96247588,"identity":"5dc54e56-fdea-4518-a38c-fa6bd5ed0874","added_by":"auto","created_at":"2025-11-19 07:27:35","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4020041,"visible":true,"origin":"","legend":"","description":"","filename":"SWCManuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/4d8710d7a3f8ab044a3ee6aa.docx"},{"id":96246452,"identity":"d70569d0-7091-4403-8c7d-1a15fad8469d","added_by":"auto","created_at":"2025-11-19 07:26:03","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5286,"visible":true,"origin":"","legend":"","description":"","filename":"1280135f82214934b63c10cd49e4327b.json","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/f09b923d1afc6757d559d975.json"},{"id":96065966,"identity":"30fc0eec-7264-4f88-a930-7bfb3bc11e43","added_by":"auto","created_at":"2025-11-17 09:19:17","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":294037,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFiles.docx","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/96b9f32d3624b6291327ab9d.docx"},{"id":96065959,"identity":"cc1a1698-2519-475b-a65a-14dae1bce7e3","added_by":"auto","created_at":"2025-11-17 09:19:17","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":179141,"visible":true,"origin":"","legend":"","description":"","filename":"1280135f82214934b63c10cd49e4327b1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/825daa94557772cee7aeca8d.xml"},{"id":96248478,"identity":"f3b18b7a-dd25-437a-b01e-8ccbc7e851e4","added_by":"auto","created_at":"2025-11-19 07:28:29","extension":"eps","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":136573898,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage2.eps","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/db79b568713c0d20750f5d44.eps"},{"id":96066009,"identity":"b4a680b6-c96c-4a8e-b0f2-542aabc39b42","added_by":"auto","created_at":"2025-11-17 09:19:18","extension":"eps","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":135819846,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage3.eps","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/ab97ee543f640b23e1dcf4c0.eps"},{"id":96247231,"identity":"2a566a04-1b53-4b94-8ed2-df4837f03e07","added_by":"auto","created_at":"2025-11-19 07:27:16","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":143562,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/8f5897c66a11f1794f8e9e89.png"},{"id":96248623,"identity":"63e9fa26-c858-4dec-ac04-00d64093749f","added_by":"auto","created_at":"2025-11-19 07:28:49","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":32483,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage10.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/0c6ce7db164a0ed87bc73c2d.png"},{"id":96065985,"identity":"4b036797-a5cd-4113-82d0-560ee79c9807","added_by":"auto","created_at":"2025-11-17 09:19:18","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":31645,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage11.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/2feef69279c8f9070085408d.png"},{"id":96248264,"identity":"3efbc04d-98cc-4ce8-a411-4fd330fc762c","added_by":"auto","created_at":"2025-11-19 07:28:14","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":45009,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage12.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/de15d34fa2808d57466101e1.png"},{"id":96065972,"identity":"9f5fcecc-a5b0-4d5a-a5b5-ed0cbdbc56da","added_by":"auto","created_at":"2025-11-17 09:19:17","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":49510,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage13.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/452bbeb9e459f26792319626.png"},{"id":96065970,"identity":"cd645276-9193-4862-9692-5ea52089667b","added_by":"auto","created_at":"2025-11-17 09:19:17","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":40281,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage14.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/56de5cacf59b4e89ff0565c9.png"},{"id":96246658,"identity":"96a3dcc1-a575-44f3-b8e5-d975414b6ddb","added_by":"auto","created_at":"2025-11-19 07:26:27","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":51592,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage15.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/e09a5bd75ea683ca4484c11f.png"},{"id":96065980,"identity":"ac1f7991-245b-4cf8-9ca4-849719b82fc6","added_by":"auto","created_at":"2025-11-17 09:19:17","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":44585,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage16.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/d0b327186931b465a5de925a.png"},{"id":96247548,"identity":"7b44d8c6-cb6c-4f95-80e8-b91eef20c74a","added_by":"auto","created_at":"2025-11-19 07:27:33","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":30060,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage17.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/b0afcb402fc01eb7d6837178.png"},{"id":96248738,"identity":"00007de5-5139-443b-a14d-421e0df66997","added_by":"auto","created_at":"2025-11-19 07:29:05","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":183583,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/b05ef8566c22200e9b545758.png"},{"id":96065982,"identity":"22595efe-d58d-44ce-afd5-3352b01626b3","added_by":"auto","created_at":"2025-11-17 09:19:17","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":8751,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/cab4c000daa24e6255e36d97.png"},{"id":96065976,"identity":"db247465-7ddf-4f8a-befb-1a8d70d399c2","added_by":"auto","created_at":"2025-11-17 09:19:17","extension":"png","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":21035,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/6005d2ca077c166fe42cc5cb.png"},{"id":96065977,"identity":"d245352f-d95d-4e1c-99b7-71f6d9b3b34c","added_by":"auto","created_at":"2025-11-17 09:19:17","extension":"png","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":43237,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/ffca1d9f965dd63c2b2840ae.png"},{"id":96065989,"identity":"d9e4c8a9-bbd4-46e2-bd5f-573f4ca53f1e","added_by":"auto","created_at":"2025-11-17 09:19:18","extension":"png","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13154,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/7ee1675843020ca1a07faa83.png"},{"id":96246662,"identity":"db4d11d5-01af-4c9c-bbde-90a1497609fc","added_by":"auto","created_at":"2025-11-19 07:26:27","extension":"png","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":262728,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/aeaaa0f242949358dbdf591b.png"},{"id":96065994,"identity":"db447c66-d97d-4fbc-aa11-813440671add","added_by":"auto","created_at":"2025-11-17 09:19:18","extension":"png","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":222688,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/d16b023ffb37b881195d9ffc.png"},{"id":96065984,"identity":"8f48dfce-2c57-4c01-bba4-17ec4cebcd42","added_by":"auto","created_at":"2025-11-17 09:19:18","extension":"png","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":50053,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage9.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/d555be6a302ec7f5be26b13a.png"},{"id":96065986,"identity":"5c5c21f3-312e-41e8-83ab-0a855ece34e2","added_by":"auto","created_at":"2025-11-17 09:19:18","extension":"jpeg","order_by":25,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":48430,"visible":true,"origin":"","legend":"","description":"","filename":"groupimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/f68d1da0f5521ba926edc011.jpeg"},{"id":96248691,"identity":"e2d4cb31-3ac8-499d-9e7c-9b5fa4f884e8","added_by":"auto","created_at":"2025-11-19 07:28:59","extension":"png","order_by":26,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":23679,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/9fad78f5e94fcac79c83d403.png"},{"id":96246727,"identity":"16aeab1b-c93d-48b6-bfe8-dde4dc6ebf37","added_by":"auto","created_at":"2025-11-19 07:26:36","extension":"png","order_by":27,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4860,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage10.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/c406c5c8c2f4794dcbf4414e.png"},{"id":96066004,"identity":"ee6308b3-0ab1-47b9-aadf-9554c55619a2","added_by":"auto","created_at":"2025-11-17 09:19:18","extension":"png","order_by":28,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4760,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage11.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/dbef05ef0877725681757b31.png"},{"id":96065987,"identity":"f329ea3c-c209-47e0-aa74-5c169df2b160","added_by":"auto","created_at":"2025-11-17 09:19:18","extension":"png","order_by":29,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6116,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage12.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/2a13453ff803c9fa2286c41f.png"},{"id":96249094,"identity":"ebe53b86-faca-4cb4-8bcf-f213fd48aeea","added_by":"auto","created_at":"2025-11-19 07:30:15","extension":"png","order_by":30,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6526,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage13.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/683a9db18e32057451e49b9b.png"},{"id":96249093,"identity":"7705b137-0ad2-4f66-a409-7b621ae40d9f","added_by":"auto","created_at":"2025-11-19 07:30:15","extension":"png","order_by":31,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5513,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage14.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/b5e61f0e00dd2b85fb0671ba.png"},{"id":96065993,"identity":"d251df63-84e8-4f56-92c2-3e4f129658ff","added_by":"auto","created_at":"2025-11-17 09:19:18","extension":"png","order_by":32,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6599,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage15.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/ef67f527bb30c4043fb9fadc.png"},{"id":96065999,"identity":"1bb0afd9-1a40-4dd6-a33b-9fec70b6a750","added_by":"auto","created_at":"2025-11-17 09:19:18","extension":"png","order_by":33,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6073,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage16.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/b4024add2dd041644f20086a.png"},{"id":96065988,"identity":"967df320-7d67-46be-aec5-fb68eca2025c","added_by":"auto","created_at":"2025-11-17 09:19:18","extension":"png","order_by":34,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6995,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage17.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/2b1a9e0f180fa86a22b9e0f0.png"},{"id":96247910,"identity":"ab55141f-1980-41e9-ab3b-782f29fdf3c4","added_by":"auto","created_at":"2025-11-19 07:27:50","extension":"png","order_by":35,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":49503,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/621e358df794139e748d337e.png"},{"id":96246740,"identity":"bdbe7ffd-0260-498b-ab56-ec36983f0934","added_by":"auto","created_at":"2025-11-19 07:26:36","extension":"png","order_by":36,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":8042,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/f7c8af71068760a2d7d2a65e.png"},{"id":96247329,"identity":"2e394fb8-2998-4ac3-aabb-eb284405ca16","added_by":"auto","created_at":"2025-11-19 07:27:23","extension":"png","order_by":37,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10059,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/377bcafb20286b00328f9edc.png"},{"id":96249286,"identity":"a2363909-96cc-4f29-903f-a18a553aec0d","added_by":"auto","created_at":"2025-11-19 07:32:51","extension":"png","order_by":38,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":28302,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/04521f2c45e994932eb69bbb.png"},{"id":96065998,"identity":"274a30f2-9da0-4aaf-b171-e650266f59c7","added_by":"auto","created_at":"2025-11-17 09:19:18","extension":"png","order_by":39,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":8987,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/d6b55e9298eeff21079747de.png"},{"id":96247366,"identity":"9b5f823f-7bd7-4ede-a386-6b5752ae928d","added_by":"auto","created_at":"2025-11-19 07:27:25","extension":"png","order_by":40,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":30524,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/bf02443a668b1ad9a0518d6b.png"},{"id":96247729,"identity":"31c69083-4d85-443b-9f2a-706949835a4e","added_by":"auto","created_at":"2025-11-19 07:27:41","extension":"png","order_by":41,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":35291,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/11eeed237b2fc5f212563fa8.png"},{"id":96066001,"identity":"ff8d33b7-3e45-405a-8b97-9e7e997e790c","added_by":"auto","created_at":"2025-11-17 09:19:18","extension":"png","order_by":42,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6543,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage9.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/96e81cf19f8d5d3365665459.png"},{"id":96066007,"identity":"ca953f9b-b69a-4674-914b-d7c0af852ca2","added_by":"auto","created_at":"2025-11-17 09:19:18","extension":"png","order_by":43,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":14763,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinegroupimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/d0ce82fa75ed709e758c6524.png"},{"id":96246245,"identity":"20777e21-014a-4f95-b299-d6015533eef8","added_by":"auto","created_at":"2025-11-19 07:25:10","extension":"xml","order_by":44,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":179518,"visible":true,"origin":"","legend":"","description":"","filename":"1280135f82214934b63c10cd49e4327b1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/b64549006a9ae2484fb05dba.xml"},{"id":96066005,"identity":"8d16a670-049c-4159-8011-ad7284f13b22","added_by":"auto","created_at":"2025-11-17 09:19:18","extension":"html","order_by":45,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":190286,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/77931120bd3d00a4fb4c5693.html"},{"id":96248439,"identity":"18bcfd89-abd4-4cf0-9399-9648a16be334","added_by":"auto","created_at":"2025-11-19 07:28:27","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":140151,"visible":true,"origin":"","legend":"\u003cp\u003eThe proposed autonomous knowledge-based smart waste collection system.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/ac1a8c60b48f3d7ab8b6fa3b.png"},{"id":96247696,"identity":"26f39ace-4852-4415-bd90-1d37df26fd6e","added_by":"auto","created_at":"2025-11-19 07:27:41","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":67113,"visible":true,"origin":"","legend":"\u003cp\u003e: Street map of the study area and bin locations.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/711ea355ee035abe33d623b3.png"},{"id":96065954,"identity":"5cc416a8-e93e-412a-904b-8febbde33634","added_by":"auto","created_at":"2025-11-17 09:19:17","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":34263,"visible":true,"origin":"","legend":"\u003cp\u003eFramework of the AI-based waste generation prediction.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/77f81ae8132ce071acfbba13.png"},{"id":96065960,"identity":"1966ad76-57cd-4c34-8c01-556f97bedc38","added_by":"auto","created_at":"2025-11-17 09:19:17","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":38001,"visible":true,"origin":"","legend":"\u003cp\u003eBin-selection algorithm for the examined scenarios.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/e21a7c235e30df46ffa26eb9.png"},{"id":96065956,"identity":"c87f0ddb-5da5-4fe6-bd63-bb3b7290e2f9","added_by":"auto","created_at":"2025-11-17 09:19:17","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":74851,"visible":true,"origin":"","legend":"\u003cp\u003eDaily waste generation pattern in the study area during the analysis period.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/ecd7c6fa7a16ffa4d0663ba3.png"},{"id":96065962,"identity":"d65d01fe-51a5-4223-a52b-5115fc0106b5","added_by":"auto","created_at":"2025-11-17 09:19:17","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":74762,"visible":true,"origin":"","legend":"\u003cp\u003eComparison between the actual and predicted daily waste generation per bin using a) the GLM and b) XGBoost.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/b536c4dcb2b7dee3f1ecbbf8.png"},{"id":96247571,"identity":"030642a2-9ec9-4c9e-99f6-d4e4ffb18869","added_by":"auto","created_at":"2025-11-19 07:27:34","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":187609,"visible":true,"origin":"","legend":"\u003cp\u003eDaily patterns of a) the amount of collected waste and b) the number of bins serviced for the examined scenarios.\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/72fd61936244922faf3ef0df.png"},{"id":96248770,"identity":"e4be0300-9ff2-400a-9a87-717d39ecfd6a","added_by":"auto","created_at":"2025-11-19 07:29:09","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":55771,"visible":true,"origin":"","legend":"\u003cp\u003eDaily collection truck travel distance patterns for the examined scenarios.\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/b3abb240f83a7b4820cb3455.png"},{"id":96065967,"identity":"8836d4a6-8168-4d78-861d-10d0181349b0","added_by":"auto","created_at":"2025-11-17 09:19:17","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":259828,"visible":true,"origin":"","legend":"\u003cp\u003eOptimized Routes of the a) CWC, b) SWC (Low-service Day), c) SWC (High-service Day), d) KWC (Low-service Day), and e) KWC (High-service Day)\u003c/p\u003e","description":"","filename":"9.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/633cee9d654ec05af905783d.png"},{"id":96246976,"identity":"2ac1b5bb-5755-47ae-b6f6-3f712ffc1aa7","added_by":"auto","created_at":"2025-11-19 07:26:58","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":106199,"visible":true,"origin":"","legend":"\u003cp\u003eSensitivity analysis on various measures for a) traffic volume and b) operating speed.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eCWC: Conventional Waste Collection, SWC: \u0026nbsp;\u0026nbsp;Sensors-based Waste Collection, KWC: Knowledge-based Waste Collection, \u0026nbsp;\u0026nbsp;CWC-CAV: Conventional Autonomous Waste Collection, SWC-CAV: Sensor-based \u0026nbsp;\u0026nbsp;Autonomous Waste Collection, KWC-CAV: Knowledge-based Autonomous Waste \u0026nbsp;\u0026nbsp;Collection\u003c/em\u003e\u003c/p\u003e","description":"","filename":"10.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/3598bedc2ddc8a53e4612bd4.png"},{"id":96065969,"identity":"dfd8250a-d1c9-4840-9d58-2e037fb1e54e","added_by":"auto","created_at":"2025-11-17 09:19:17","extension":"png","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":245670,"visible":true,"origin":"","legend":"\u003cp\u003eHeatmap of the traffic volume in the most utilized streets for a) CWC, b) SWC (low-service day), c) SWC (high-service day), d) KWC (low-service day), and e) KWC (high-service day).\u003c/p\u003e","description":"","filename":"11.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/1a600f9c7410968536f304aa.png"},{"id":96065963,"identity":"a4a285c1-742d-4da0-8a6f-d1747f5d9fbf","added_by":"auto","created_at":"2025-11-17 09:19:17","extension":"png","order_by":12,"title":"Figure 12","display":"","copyAsset":false,"role":"figure","size":55799,"visible":true,"origin":"","legend":"\u003cp\u003eAnnual net present cost of the examined scenarios.\u003c/p\u003e","description":"","filename":"12.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/a99dda08bdf5624478268939.png"},{"id":96248743,"identity":"1c6bfd42-0831-4246-aa05-df0be99c133a","added_by":"auto","created_at":"2025-11-19 07:29:05","extension":"png","order_by":13,"title":"Figure 13","display":"","copyAsset":false,"role":"figure","size":577601,"visible":true,"origin":"","legend":"\u003cp\u003ePercentage reduction in the a) number of serviced bins, b) number of collection days, c) amount of collected waste, d) travel distance, e) travel time, f) queue length, g) traffic delay, h) carbon footprint, and i) net present cost for the examined scenarios compared to the conventional waste collection system.\u003c/p\u003e","description":"","filename":"13.png","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/9762b708b55b9a4b7b5dda5e.png"},{"id":107928703,"identity":"695190e9-9a60-414b-9cc7-75c00b9fe47b","added_by":"auto","created_at":"2026-04-27 16:12:05","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2487557,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/430d1ae0-65f3-46ae-8bbc-2d78895d4bb0.pdf"},{"id":96247254,"identity":"3db7e54b-c858-4233-a7db-9a20871e4ffb","added_by":"auto","created_at":"2025-11-19 07:27:18","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":294037,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFiles.docx","url":"https://assets-eu.researchsquare.com/files/rs-8012685/v1/b0a4c55872abbab83e39f36e.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Development and Simulation of an Innovative Autonomous Knowledge-Based Smart Waste Collection System","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eSolid waste management is one of the major challenges faced by municipalities worldwide due to its direct impact on the environment and public health, as well as its economic burden and extensive operation. The rapid growth in waste generation rates has led to more challenging management processes, particularly in the collection system. The conventional waste collection approach involves a fleet of trucks transporting waste collected from bins to intermediate or final disposal sites. This process leads to ongoing issues, including air pollution and traffic congestion, and typically accounts for more than half the total waste management costs. Moreover, the daily and seasonal variations in waste generation rates result in low efficiency of the traditional predefined fixed collection route regardless of the actual bin use capacity (Abdallah et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). This results in the frequent servicing of waste bins that are only partially full at the time of collection (Ramos et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). The socioeconomic and environmental implications of such inefficient operation suggest that smart collection systems that respond to the actual bin use will provide potential improvements.\u003c/p\u003e\u003cp\u003eRecent efforts have focused on modernizing this inefficient process by developing sensor-based smart waste collection systems, aiming to reduce the frequency of waste collection events and consequently lower fuel consumption, carbon emissions, and traffic congestion. Smart waste collection can be categorized into two areas: 1) data acquisition and communication hardware, particularly at the waste bin, and 2) decision-making and route optimization algorithms, along with simulations under various operating scenarios. Overall, compared to conventional systems, previous studies reported theoretical improvements in operation and maintenance costs, total travel time, fuel consumption, and air pollution emissions. Most hardware studies focused on installing sensors in waste bins (Mamun, Hannan, Hussain, \u0026amp; Basri, 2016); (Gutierrez, Jensen, Henius, \u0026amp; Riaz, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2015\u003c/span\u003e); (Catania \u0026amp; Ventura, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2014\u003c/span\u003e); (Juwariyah et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2021\u003c/span\u003e); (Sidhu et al., \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), whereas studies that considered onboard truck sensors are lacking. However, several challenges limit the practical feasibility and scalability of these systems, including heavy reliance on sensor instrumentation, periodic maintenance, significant upfront costs, calibration, signal interference, deliberate or unintentional tampering, and power supply issues, among others.\u003c/p\u003e\u003cp\u003eCollection trucks equipped with onboard weight sensors may offer several advantages compared to sensor-equipped bins, such as 1) elimination of the required wireless communications between bins and a control center, 2) fewer monitored vehicles compared to numerous waste bins, 3) easier centralized maintenance at the garage versus scattered service points (bins), 4) longer sensor lifetime due to less indirect contact with waste and continuous extreme weather conditions, and 5) security of the installed sensors due to inaccessibility to the public and protection against vandalism, all of which lead to a more robust, practical, simple, and cost-effective operation. This study replaces sensor-equipped bins with weight sensors onboard trucks that record the amount of daily generated waste at the bin level. Additionally, the amount of waste per bin can be forecasted for the next day using the recorded quantities, allowing for optimized collection routes for bins that require servicing.\u003c/p\u003e"},{"header":"2. Literature Review","content":"\u003cp\u003eThe developed knowledge-based waste collection system integrates daily waste generation forecasting, bin-selection decision-making, route optimization, and autonomous transportation. The following section provides a comprehensive literature review, examining each component as a standalone technology and as an integral part of a dynamic waste collection system.\u003c/p\u003e\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1. Waste Generation Forecasting\u003c/h2\u003e\u003cp\u003eSeveral studies have forecasted waste generation with varying temporal and spatial resolutions, where most of the literature is focused on predicting the annual (Jassim et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Lu et al., \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Latif et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Srivastava \u0026amp; Jha, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2023\u003c/span\u003e) and monthly (Abbasi et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Fan et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Ghanbari et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Liang et al., \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) produced quantities in cities and countries. Multiple studies have predicted the amount of solid waste generated in cities, varying the temporal resolution from weekly to daily, as summarized in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. For instance, the weekly household waste (HSW) in nine regions Southeast of the United Kingdom was forecasted using a proposed ensemble meta regressor, which applied linear regression (LR) as a meta learner to be trained on the predicted optimized outcome of various machine learning models (Namoun et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). The meta regressor outperformed all the individual models, achieving a coefficient of determination (R\u0026sup2;) of 0.80. The weekly refuse and recyclable waste were forecasted in 609 sub-sections of New York City using a Gradient-boosting regression tree, which achieved an R\u003csup\u003e2\u003c/sup\u003e of 0.87 (Kontokosta et al., \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2018\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eSummary of the best-performing machine learning models forecasting waste generation in the literature.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"7\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eReference\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eModel\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eTemporal Resolution\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eSpatial Resolution\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eWaste Type\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eFactors\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c7\"\u003e\u003cp\u003eR\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGuha et al. (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2023\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMultiple linear regression\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eAverage daily\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e381 households in 31 wards\u003c/p\u003e\u003cp\u003ein Rangpur, Bangladesh\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eHSW\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.66\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAdusei et al. (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2022\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eRecurrent neural network\u003c/p\u003e\u003cp\u003elong short-term memory\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eDaily (2013\u0026ndash;2021)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eCity of Regina Landfill, Canada\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eMSW\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.72\u0026ndash;0.86\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMudannayake et al. (\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2022\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eRandom forest\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eDaily (2011\u0026ndash;2013)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eBallart, Australia (2001\u0026ndash;2011)\u003c/p\u003e\u003cp\u003eAustin, USA (2005\u0026ndash;2019)\u003c/p\u003e\u003cp\u003eBoralesgamuwa, Sir Lanka (2012\u0026ndash;2018)\u003c/p\u003e\u003cp\u003eDehiwala, Sir Lanka (2012\u0026ndash;2015)\u003c/p\u003e\u003cp\u003eMoratuwa, Sir Lanka (2015\u0026ndash;2018)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eMSW\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eTimeseries\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003eMAPE:\u003c/p\u003e\u003cp\u003eBallart: 8.3%\u003c/p\u003e\u003cp\u003eAustin: 9.0%\u003c/p\u003e\u003cp\u003eBoralesgamuwa: 29.5%\u003c/p\u003e\u003cp\u003eDehiwala: 36.7%\u003c/p\u003e\u003cp\u003eMoratuwa: 33.1%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNamoun et al. (\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2022\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eEnsemble meta regressor\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eWeekly (2011\u0026ndash;2021)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e9 regions Southeast UK\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eHSW\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.80\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAraiza-Aguilar et al. (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2020\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMultiple linear regression\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eAverage daily\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e118 localities in Cuenca del\u003c/p\u003e\u003cp\u003eCa\u0026ntilde;\u0026oacute;n del Sumidero, Mexico\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eMSW\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.975\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNoufal et al. (\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2020\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eMultiple linear regression\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePer capita average daily\u003c/p\u003e\u003cp\u003e(2 weeks in Jul 2017, Aug 2018, Jan \u0026amp; Feb 2019)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e300 households in Homs, Syria\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eHSW\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003ePer capita: 0.56\u003c/p\u003e\u003cp\u003eRecyclable: 0.71\u003c/p\u003e\u003cp\u003eOrganics: 0.84\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eKontokosta et al. (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2018\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eGradient boosting regression tree\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eWeekly (2013\u0026ndash;2016)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e609 sub-sections in New York City, USA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eRefuse and\u003c/p\u003e\u003cp\u003erecyclables\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e31\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.87\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHoang et al. (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2017\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eBayesian model average method\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eAverage daily\u003c/p\u003e\u003cp\u003e(2 weeks in 2015)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e286 households in Hoi An, Vietnam\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eHSW\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.33\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSong et al. (\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2014\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSimulated annealing-based variable weighted combining the chaotic model, ANN, and PLS-SVM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eDaily (2011\u0026ndash;2013)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eSeattle, USA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eMSW\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eTimeseries\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.94\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eThis study\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eGeneralized linear model and gradient-boosting trees\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eDaily (2019\u0026ndash;2020)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eAbu Dhabi, UAE\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eRecyclables\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.94\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"7\" nameend=\"c7\" namest=\"c1\"\u003e\u003cp\u003eANN: artificial neural networks, PLS-SVM: partial least square support vector machine, HSW: household solid waste, MSW: municipal solid waste, MAPE: mean absolute percentage error\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eOn the other hand, the daily municipal solid waste (MSW) generation was forecasted in five cities across Australia, the USA, and Sri Lanka, spanning different temporal durations (Mudannayake et al., \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Various timeseries and deep learning models were utilized based on single- and multi-model prediction approaches. Random forest showed the best performance with a mean absolute percentage error ranging from 8.3% to 36.7% using the single model forecasting method. Song et al. (\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2014\u003c/span\u003e) predicted daily MSW using a timeseries simulated annealing hybrid model, which combines and weighs the outputs of various models. The proposed hybrid model had the best performance, with an R\u003csup\u003e2\u003c/sup\u003e of 0.94 (Song et al., \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). In addition, the daily MSW was forecasted by a recurrent neural network long short-term memory model in Saskatchewan, Canada, with an R\u003csup\u003e2\u003c/sup\u003e ranging from 0.72 to 0.86 for different meteorological seasons (Adusei et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eOther studies have focused on predicting daily waste generation at varying spatial resolutions, ranging from households to divisions within a city. The Bayesian model average method was employed to predict the average HSW, surveyed across two weeks from 286 households in Hoi An, Vietnam (Hoang et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). In addition, multiple linear regression (MLR) was implemented by Guha et al. (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2023\u003c/span\u003e) and Araiza-Aguilar et al. (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2020\u003c/span\u003e), in which the former predicted the average HSW for 381 households across 31 wards in Rangpur City Corporation, Bangladesh, to identify suitable landfill sites, whereas the latter forecasted the average daily MSW for 118 localities in Cuenca del Ca\u0026ntilde;\u0026oacute;n del Sumidero, Mexico, with an R\u003csup\u003e2\u003c/sup\u003e of 0.975 (Araiza-Aguilar et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Guha et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). MLR was also utilized to forecast the per capita average daily HSW, as well as the daily fraction of recyclables and organics for 300 households in Homs, Syria, where the prediction of organic waste achieved the highest R\u003csup\u003e2\u003c/sup\u003e of 0.84 (Noufal et al., \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e2.2. Route Optimization\u003c/h2\u003e\u003cp\u003eMost of the literature applied route optimization for waste collection at garbage accumulation points (Mah\u0026eacute;o et al., \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), dumping sites (Rambandara et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), central collection points/gather sites (Amal et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Hu et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Zhang et al., \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), separation centers (Mohammadi et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), and districts (Greco et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). This is due to the adapted fixed routes in local neighborhoods, which are optimized in terms of travel distance and time. However, dynamic routing systems should continuously adapt due to daily variations in the amount of waste per bin, requiring ongoing optimization to efficiently service fluctuating needs (Alsobky et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Mohammadi et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eSome studies have optimized the MSW collection route at the bin level using various methods. For instance, seven scenarios were proposed to optimize the travel distance and collection time for a neighborhood in Cairo, Egypt, by applying an integer genetic algorithm to solve the traditional salesman problem (Alsobky et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). The baseline collection scenario consisted of 235 bins with a volume of 240 liters of unsorted MSW serviced once daily by medium- and five small pick-up vehicles with manual loading and no compaction. In contrast, the best-performing scenario involved a medium truck with mechanical loading and compaction, collecting sorted waste from 73 collection points through two daily rounds, which reduced the travel distance and collection time by 70% and 57%, respectively. Another study developed a hybrid particle swarm optimization (PSO) and firefly algorithm (FA) using local search (I-HFPSO) to optimize the waste collection route of 280 containers in a district in Sanlıurfa, Turkey (Kaya, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). I-HFPSO outperformed linear programming, ArcGIS, FA, PSO, and HFPSO, resulting in a 31.1% reduction in travel distance compared to the current practice.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e2.3. Bin-Selection Decision-Making\u003c/h2\u003e\u003cp\u003eFew studies have implemented dynamic waste collection, which accounts for the varying daily amounts of waste and optimizes collection routes accordingly, as summarized in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e2\u003c/span\u003e. For instance, a dynamic waste collection algorithm was proposed that employs a look-ahead heuristic to select bins that are full on the current collection day and those expected to reach capacity on the next collection day, which may not necessarily align with the subsequent calendar day, based on statistically estimated daily waste accumulation rates per bin (Jorge et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). A simulated annealing (SA)/neighborhood search (NS) algorithm determines additional profitable bins to collect and optimizes the collection route of the selected bins. The hybrid metaheuristic was compared to actual operations in Portugal, where different runs yielded a profit ranging between 45% and 60%. Additionally, an algorithm for bin selection and route optimization was developed using a mixed-integer linear programming model and tested on both small, randomly generated instances and large instances adapted from an existing split delivery vehicle routing problem (Luo et al., \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). The developed system utilizes the variable NS framework and employs SA as the acceptance criterion. Sites for waste collection were selected based on whether they exceeded a set threshold or were predicted to overflow the next day. The system allowed multiple vehicles to service a site if its waste surpassed the capacity of one vehicle. The performance of the proposed model using small-scale instances was compared to four competing state-of-the-art algorithms and CPLEX, showing a reduction in operating costs and fleet size by approximately 50% when implementing an 80% collection threshold, effectively preventing overflow. Moreover, a multi-agent simulation-based modeling approach was proposed, comprising agents representing households, planners, and drivers (Hussain et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Waste density was determined in real-time by wireless sensors that transmit data to a cloud platform. This data was then monitored by the planner agent, who decides whether to initiate a waste collection trip based on the number of bins exceeding a predetermined threshold. The wireless sensor-based scenario achieved significantly reduced travel distances, carbon dioxide emissions, collection times, the number of overfilled bins, and the number of trucks. Furthermore, a knowledge-based dynamic waste collection system was developed for a residential area (Abdallah et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). The economic and environmental benefits were estimated to reach 40% through analysis of waste generation data, a bin-selection algorithm, and geographic information system (GIS) route optimization. A smart system entirely dependent on hardware was considered impractical under the harsh and aggressive operating conditions of the costly waste collection process.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eSummary of studies that implemented dynamic waste collection systems, including bin selection and route optimization.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"11\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c11\" colnum=\"11\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAuthors\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eData source\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eWaste type\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eMonitored factors\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eMonitoring\u003c/p\u003e\u003cp\u003emethod\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eBin selection\u003c/p\u003e\u003cp\u003econditions\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c7\"\u003e\u003cp\u003eRoute optimization constraints\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c8\"\u003e\u003cp\u003eObjective\u003c/p\u003e\u003cp\u003efunction\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c9\"\u003e\u003cp\u003eBin capacity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c10\"\u003e\u003cp\u003eCapacity threshold\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c11\"\u003e\u003cp\u003eFleet size\u003c/p\u003e\u003cp\u003ecapacity\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e(Luo et al., \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2024\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSimulated data for community collection sites\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eLevel\u003c/p\u003e\u003cp\u003eWaste generation quantity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eInitially known (unspecified)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u0026gt; threshold\u003c/p\u003e\u003cp\u003e(collection day)\u003c/p\u003e\u003cp\u003e\u0026thinsp;\u0026gt;\u0026thinsp;bin capacity\u003c/p\u003e\u003cp\u003e(next day)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u0026lt; vehicle capacity\u003c/p\u003e\u003cp\u003eMultiple vehicles can collect\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003eMin cost\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e\u003cp\u003eSmall instances:\u003c/p\u003e\u003cp\u003e0\u0026ndash;2Q\u003c/p\u003e\u003cp\u003eLarge instances:\u003c/p\u003e\u003cp\u003e0 - (1.5\u0026times;I0i)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e\u003cp\u003eSmall instances:\u003c/p\u003e\u003cp\u003e0%, 60%, 100%\u003c/p\u003e\u003cp\u003eLarge instances:\u003c/p\u003e\u003cp\u003e0, 60% \u0026minus;\u0026thinsp;100%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e\u003cp\u003e2\u0026ndash;41\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e(Jorge et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2022\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSimulated data\u003c/p\u003e\u003cp\u003eReal data (Lisbon, Portugal)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePaper\u003c/p\u003e\u003cp\u003eCardboard\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eLevel\u003c/p\u003e\u003cp\u003eAccumulation rate\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eReal-time waste\u003c/p\u003e\u003cp\u003eLevel sensors\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u0026ge; bin capacity\u003c/p\u003e\u003cp\u003e(collection day)\u003c/p\u003e\u003cp\u003e\u0026thinsp;\u0026ge;\u0026thinsp;bin capacity\u003c/p\u003e\u003cp\u003e(next collection day)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u0026lt; vehicle capacity\u003c/p\u003e\u003cp\u003eMaximum labor shift duration\u003c/p\u003e\u003cp\u003eMaximum route workload balance\u003c/p\u003e\u003cp\u003eRemoval of arc intersections\u003c/p\u003e\u003cp\u003eModel running time\u0026thinsp;\u0026le;\u0026thinsp;2 hours\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003eMax profit\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e\u003cp\u003e74 kg\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e\u003cp\u003e100%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e\u003cp\u003e1\u0026ndash;2**\u003c/p\u003e\u003cp\u003e4 tons\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e(Hussain et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2024\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSimulated data \u003c/p\u003e\u003cp\u003e(Al Rayyan, Qatar)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eHousehold\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eDensity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eReal-time sensors\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u0026gt; threshold\u003c/p\u003e\u003cp\u003e\u0026thinsp;\u0026gt;\u0026thinsp;bin capacity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u0026lt; vehicle capacity\u003c/p\u003e\u003cp\u003eEach bin collected by one vehicle\u003c/p\u003e\u003cp\u003e1 vehicle can collect multiple rounds\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003eMin distance\u003c/p\u003e\u003cp\u003eMin cost\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e\u003cp\u003e1 m\u003csup\u003e3\u003c/sup\u003e\u003c/p\u003e\u003cp\u003e312 kg/m\u003csup\u003e3\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e\u003cp\u003e60%, 70%,\u003c/p\u003e\u003cp\u003e80%, 90%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e(Martikkala et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2023\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSimulated data based on sensors (Seinajoki, Finland)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eTextile\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eLevel\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eReal-time sensors\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u0026gt; threshold\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003eCapacity/time constraint\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003eMin cost\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e\u003cp\u003e0.24 m\u003csup\u003e3\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e\u003cp\u003e90% (previous evening)\u003c/p\u003e\u003cp\u003e80% (collection day)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e(Belhiah et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2024\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eReal data (Tangier, Morocco)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eGreen\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eReal-time sensors\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003eMin distance\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e\u003cp\u003e300\u003c/p\u003e\u003cp\u003emanual sweepers\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e(Brouwer et al., \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2023\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eReal data (ERSUC, Coimbra, Portugal)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e2 Paper\u003c/p\u003e\u003cp\u003e1 Plastic/metal\u003c/p\u003e\u003cp\u003e1 Glass\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eLevel\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eReal-time sensors\u003c/p\u003e\u003cp\u003eVisual observation\u003c/p\u003e\u003cp\u003eMobile sensors\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e\u003cp\u003e2.5 m\u003csup\u003e3\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eThis study\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eReal data (Abu Dhabi, UAE)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eRecyclables\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eWeight\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003ePredicted\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u0026gt; threshold\u003c/p\u003e\u003cp\u003e(collection day)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003eMin distance\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c9\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c10\"\u003e\u003cp\u003e90%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c11\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"11\" nameend=\"c11\" namest=\"c1\"\u003e\u003cp\u003e**Real data scenario, I0i: the initial amount of waste at customer \u0026#119894; (tons), Q: Vehicle capacity, Recyclables: paper, plastic, metal, and glass, UAE: United Arab Emirates\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003econtinued\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"7\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAuthors\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eRoute optimization number of days\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eNumber of bins\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eUtilized\u003c/p\u003e\u003cp\u003emethods\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eOptimization\u003c/p\u003e\u003cp\u003eproblem\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eScenarios\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c7\"\u003e\u003cp\u003eBAU Collection\u003c/p\u003e\u003cp\u003eFrequency\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eLuo et al. (\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2024\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e5\u0026ndash;10\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eMILR\u003c/p\u003e\u003cp\u003eVNS\u003c/p\u003e\u003cp\u003eSA\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eVRP\u003c/p\u003e\u003cp\u003eSDVRP\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eVariable capacity, VRPHAS, SRC\u0026thinsp;+\u0026thinsp;VND, SplitILS, RGTS, CPLEX\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eDaily\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e27\u0026ndash;43\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eVariable capacity\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eJorge et al. (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2022\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e5\u0026ndash;10*\u003c/p\u003e\u003cp\u003e30**\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e50\u0026ndash;250*\u003c/p\u003e\u003cp\u003e226**\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eSA\u003c/p\u003e\u003cp\u003eNS\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eVRPP\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eExact branch-and-cut algorithm***\u003c/p\u003e\u003cp\u003eSANS without workload concerns*\u003c/p\u003e\u003cp\u003eSANS with workload concerns*\u003c/p\u003e\u003cp\u003eActual operations**\u003c/p\u003e\u003cp\u003eHybrid metaheuristic**\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003eVariable\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHussain et al. (\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2024\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e365\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e200\u0026ndash;500\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eFSM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eMTSP\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003ePeriodic review (BAU)\u003c/p\u003e\u003cp\u003eWireless sensors-based\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003eDaily, after-2-days,\u003c/p\u003e\u003cp\u003eafter-3-days, weekly\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMartikkala et al. (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2023\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e28\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e10\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eODL\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eConventional (BAU)\u003c/p\u003e\u003cp\u003eDynamic\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003eWeekly (all bins)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBelhiah et al. (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2024\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eContraction\u003c/p\u003e\u003cp\u003eHierarchies\u003c/p\u003e\u003cp\u003eOpen Source\u003c/p\u003e\u003cp\u003eRouting Machine\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBrouwer et al. (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2023\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eThis study\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e30\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e396\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eArcMap\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eTSP\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eConventional (BAU)\u003c/p\u003e\u003cp\u003eSensor-based collection (real-time)\u003c/p\u003e\u003cp\u003eknowledge-based collection (forecasted)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003eDaily\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"7\" nameend=\"c7\" namest=\"c1\"\u003e\u003cp\u003e*Simulated data scenario, **Real data scenario, ***Developed by Ramos et al., (\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2018\u003c/span\u003e), BAU: Business as usual, FSM: finite state machine, MILR: mixed integer linear programming, NS: neighborhood search, ODL: Open door logistics, RGTS: randomized granular tabu search, SA: simulated annealing, SRC\u0026thinsp;+\u0026thinsp;VND: Split-delivery route construction improved by variable neighborhood descent, TSP: traveling salesman problem, MTSP: multiple TSP, VNS: variable NS, VRP: vehicle routing problem, SDVRP: split delivery VRP, SplitILS: SDVRP solved by iterated local search, VRPHAS: VRP heuristic with a priori split strategies, VRPP: VRP with profits\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e2.4. Autonomous Vehicles\u003c/h2\u003e\u003cp\u003eResearchers previously introduced the idea of incorporating connected and autonomous vehicles (CAVs) into a waste collection system (Alfeo et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Autonomous and connected transportation can inform vehicles in advance about surrounding traffic and maneuvering decisions, thereby maintaining uniform speeds and minimizing sudden braking, which leads to reductions in fuel consumption, traffic congestion, and emissions. CAVs use a global positioning system (GPS) and light detection and ranging (LiDAR) to map, position, and scan the area around the vehicle. For waste collection applications, predetermined routes will significantly reduce the complexity of the driving task assigned to CAVs. This concept is crucial for determining the minimum level of automation required to perform the task. According to the Society of Automotive Engineers (SAE), all vehicles fall under one of six levels of automation, ranging from no automation (level 0) to full automation capabilities (level 5) (SAE, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Over the last decade, researchers have tested and validated the reliability of Level 3 CAVs, with most car manufacturers having reached the development stage for Level 4 automation. Additionally, to receive signals from waste bin sensors and communicate with other collection vehicles, CAVs will also be required to have vehicle-to‐vehicle (V2V) and vehicle‐to‐infrastructure (V2I) capabilities. In 2018, the Volvo Group developed autonomous waste collection vehicles with V2V capabilities and demonstrated the potential benefits of utilizing these CAVs in navigating obstacles and consistently driving to each bin location. Other automotive companies have also deployed autonomous waste collection trucks, such as eCanter SensorCollect by Mitsubishi Fuso, in Japan, Europe, and the USA.\u003c/p\u003e\u003cp\u003eHowever, the current research lacks the field application of autonomous vehicles for waste collection. The focus is mainly on three domains: 1) robots for collecting waste in water bodies, 2) robots for collecting trash from beaches and parks, and 3) simulation of autonomous vehicles for waste collection. For instance, a mobile autonomous robot for litter emptying (MARBLE) was developed and tested on three different areas with varying numbers of bins (Justo et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Routes were optimized using vehicle route problems with SA, game theory knapsack problems with smart litter bins, and knapsack problems without smart bins. Data is proposed to be communicated through a long-range, wide-area network (LoRaWAN). Another study proposed a system employing smart bins that measure the waste fill level to be collected by level 2 autonomous vehicles, where the communications were completed using a long-range wide area network (LoRaWAN), i.e., low power wide area (LPWA) instead of Wi-Fi and Ethernet (Karatagi et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003e2.5. Research Gap\u003c/h2\u003e\u003cp\u003eAfter more than a decade of research in this field, no fully functional smart waste collection system has been installed. Based on the reviewed literature to date, no study has utilized artificial intelligence and big data analytics to minimize the need for the typical hardware components of smart waste collection systems. Additionally, to the best of the authors' knowledge, the prediction of daily waste generation per bin is yet to be implemented. Moreover, driverless waste collection remains a contemporary research field that warrants significant attention, as intelligent transportation gradually emerges in smart cities. This is where a newly developed plan incorporating a fully automated system and an optimum routing strategy could have significant benefits.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003e2.6. Aim and Objectives\u003c/h2\u003e\u003cp\u003eThis research presents a novel approach to smart waste collection by integrating multiple state-of-the-art technologies, including artificial intelligence, predictive analytics, route optimization, and intelligent transportation. This study aims to develop a dynamic framework for an autonomous, knowledge-based smart waste collection system, deploy the envisioned system in a computer simulation of a residential area over a one-month period, and assess its economic feasibility using a life-cycle costing (LCC) analysis. Historical records were used to determine the daily amount of waste generated per bin, acquired from onboard truck sensors, to predict future waste generation and accordingly select the bins that require collection. Next, collection routes were optimized for waste trucks and simulated the proposed system implementation using autonomous vehicles. Such enhancements to the waste collection system will significantly improve its economics, particularly in terms of capital investments and operational and maintenance costs, thereby fostering the field implementation of smart collection systems due to improved profitability and functionality. Mitigating health risks and environmental hazards associated with inadequate waste collection services and traffic congestion would add to the social benefits of such a system. The proposed approach will provide government waste management authorities and private waste collection companies with an alternative, sustainable method to deliver waste management services in smart cities.\u003c/p\u003e\u003c/div\u003e"},{"header":"3. Methodology","content":"\u003cp\u003eThis research developed a multidisciplinary framework for an autonomous knowledge-based smart waste collection system and simulated its implementation in an actual residential area to collect municipal solid waste. The following subsections present the system architecture, simulation data description, and methodological procedures for the various system components, including waste generation prediction, heuristic bin selection, route optimization, autonomous navigation, and LCC analysis.\u003c/p\u003e\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\u003ch2\u003e3.1. System Architecture\u003c/h2\u003e\u003cp\u003eThis study aims to develop a new autonomous smart waste collection system that achieves the most efficient operation, maximum economic benefits, and minimum environmental impacts. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e shows the framework of the proposed system. The primary data utilized by the system are historical records of collected waste per bin, including bin locations, collection routes, and daily quantities measured via truck onboard weighing sensors and GPS systems. In addition, the research utilizes various influential parameters, including 1) types of households, 2) demographic statistics, 3) seasonal calendars, and 4) other local data affecting waste generation patterns. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, the study used the obtained databases to train machine learning models to forecast the daily waste amounts at the bin level while incorporating deep reinforcement methods for supervised learning. Subsequently, a bin-selection algorithm selects bins to be serviced based on the predicted waste generation and maximum bin capacity. The model feeds designated bins to a route optimization module, which receives live traffic data simultaneously. The module transmits the optimized collection route to a GPS-enabled autonomous collection truck. Simultaneously, the system fine-tunes the models and software modules based on the actual quantity of generated waste measured through an onboard weighing system.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003e3.2. Simulation Description\u003c/h2\u003e\u003cp\u003eA computer simulation was conducted to evaluate the various components of the proposed system in an actual residential community located in Abu Dhabi, the second-most populous city in the United Arab Emirates (UAE), after Dubai. Figure\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e shows a map of the street layout and bin locations in the study area. The community comprises residential, institutional, and governmental buildings, with 375 single-detached dwellings and 12 institutional and commercial buildings, all connected by a street network with a total length of approximately 14 km. The Abu Dhabi Waste Management Center (Tadweer) provided the daily amount of recyclable waste per bin, i.e., paper, plastic, metal, and glass, for 396 bins with a volume of 240 liters each. Recyclable waste was selected as the stream for which collection routes are optimized due to hot weather conditions that prevent other types of waste, such as organics, from being retained in bins for more than a day. The dataset spanned the period from January 2019 to December 2020, yielding approximately 290,000 observations. Each instance included the day of collection, the latitude and longitude position of each bin, and the amount of waste in kilograms measured by onboard truck sensors. A unique number was assigned to each bin to facilitate the application of the bin selection algorithm.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003e3.3. Proposed Systems and Scenarios\u003c/h2\u003e\u003cp\u003eThis study aims to compare existing traditional waste collection practices with sensor-based and knowledge-based smart waste collection systems. The study established three scenarios: 1) conventional waste collection (CWC), 2) sensor-based waste collection (SWC), and 3) knowledge-based waste collection (KWC). CWC represents the conventional waste collection system in which all bins are serviced daily through the same route. SWC, representing typical smart waste collection systems, employs weight sensors in the waste bins to obtain the amount of waste. Alternatively, the proposed knowledge-based system, i.e., KWC, utilizes historical records of daily generation patterns per bin, obtained from weight sensors onboard the collection trucks, to predict the produced quantities. The SWC and KWC systems collect only the bins that require service; hence, the collection route varies daily. It is expected that the SWC will be the best-performing system due to the acquired real-time measurements. However, it is an expensive system consisting of various costly components and intricate wireless connections that require multiple modifications to the city infrastructure and increased maintenance costs.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003e3.4. Waste Generation Prediction\u003c/h2\u003e\u003cp\u003eFigure \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e presents the framework of the various processes implemented to forecast the daily waste generated per bin. Data preprocessing was performed by cleaning and transforming the raw data compiled from the waste collection company. This involved removing duplicate records, handling missing or inconsistent entries, and filtering out bins with incomplete location data. Extreme outliers, primarily caused by sensor malfunctions, were identified using the interquartile range method. Moreover, categorical variables such as day of week and season were encoded into numerical features to capture temporal variation. The dataset was split into 67% and 33% subsets for training and testing, respectively. These fractions correspond to model training from January 4, 2019, until March 31, 2020, with testing starting the following day. Feature engineering, including feature generation and selection, was implemented to eliminate redundant or irrelevant features from the dataset, thereby enhancing accuracy and facilitating a better understanding of the underlying relationship between dependent and independent variables.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe study trained two machine learning algorithms, a generalized linear model (GLM) and gradient-boosted trees (XGBoost), on historical daily waste generation data per bin. A GLM is an extension of conventional linear models, combining both linear and logistic regression, in which the data are fitted by following a maximum likelihood approach (Zheng \u0026amp; Agresti, \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2000\u003c/span\u003e). It offers multiple regression advantages within a flexible framework; hence, the model-fitting computation is rapid and parallel. Moreover, a GLM is adequate for models with missing values and a limited number of nonzero parameter predictors. On the other hand, XGBoost is a set of classification or regression decision trees that employ a flexible nonlinear regression procedure, providing a forward-learning ensemble method to obtain predictive results through incrementally improved estimations. XGBoost effectively enhances accuracy due to its sensitivity to the computational range of data and features; thus, it does not require data normalization (Sun Yin \u0026amp; Vatrapu, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). These two models were selected to represent complementary approaches, where the GLM serves as a transparent baseline model that captures linear dependencies, and XGBoost is an advanced ensemble method capable of modeling nonlinear relationships. This combination enabled benchmarking a simple interpretable model against a state-of-the-art machine learning algorithm. This study utilized RapidMiner Studio software to develop the selected models.\u003c/p\u003e\u003cp\u003eEach model was trained and optimized by running multiple iterations to determine the optimal parameters. Hyperparameter optimization was carried out through grid search or parameter sweep, the most applied method for searching through a predefined subset of the learning algorithm hyperparameter space. The study specified these parameters using minimum and maximum values, i.e., lower and upper bounds, along with the number of incremental steps with different scales, e.g., linear, logarithmic, and quadratic scales. In a grid search, trials were established by aggregating every conceivable value combination. This study employed a 3-fold cross-validation (CV) and grid search optimization approach to optimize three parameters of the XGBoost model: the number of trees, maximal depth, and learning rate, with ranges of 30\u0026ndash;150, 2\u0026ndash;7, and 0.001-0.1, respectively. The optimal number of trees, maximum depth, and learning rate were found to be 150, 7, and 0.1, respectively. Afterward, a CV method was deployed to test the models. Evaluation metrics were applied to assess the performance of the developed waste prediction models, specifically the root mean squared error (RMSE), squared error (SE), absolute error (AE), relative error (RE), and R-squared (R\u003csup\u003e2\u003c/sup\u003e).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003e3.5. Bin-selection Algorithm\u003c/h2\u003e\u003cp\u003eThe three examined scenarios were modeled using the MATLAB programming platform, following the diagram shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e. The bin-selection algorithm utilized the predicted amount of waste in each bin obtained from the best-performing AI model to select the bins to be collected. A heuristic system determined whether the predicted waste volume exceeded the effective bin capacity. If so, it assigned the bin for collection. Otherwise, the system skipped the bin along the collection route and repeated the capacity check for the same bin the next day, and so on. For instance, the CWC scenario represented the traditional system that collects all the bins every day. On the other hand, the bin-selection algorithm in the SWC and KWC scenarios selected only the bins that reached 90% of their maximum capacity, i.e., considering a 10% safety margin. On any given day, the amount of waste in bins that were not collected accumulates.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\u003ch2\u003e3.6. Route Optimization\u003c/h2\u003e\u003cp\u003eThe system used the selected bins acquired from the bin-selection algorithm to optimize the route for waste collection trucks. Route optimization was implemented using street locations and the position of the selected bins in terms of latitude and longitude coordinates. The Network Analyst extension in ArcMap software was utilized to optimize routes based on the Dijkstra algorithm, which determines the shortest path in terms of total travel distance. For instance, the system considered all possible routes between bins and selected the one with the minimum length. The system performed route optimization only once to collect all bins in the CWC scenario, whereas the SWC and KWC scenarios applied route optimization daily, as the selected bins changed daily in response to variations in waste generation.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\u003ch2\u003e3.7. Autonomous Navigation\u003c/h2\u003e\u003cp\u003eThis research carried out traffic modeling and simulation using PTV Vissim software. The analysis comprises three main steps: 1) establish layout and geometry considerations for the study area, 2) compile traffic data for the major and minor roads on weekdays and weekends, and 3) conduct ten simulation runs over a month to produce a reliable estimation due to the stochastic nature of the simulation. The study randomly simulated the time spent at each bin following a normal distribution with a mean of 60 seconds and a standard deviation of 30 seconds. The waste collection truck route in the CWC, SWC, and KWC scenarios was calibrated and validated using human driving behavior to reflect the actual conditions of the network categories, as well as predicted autonomous driving behavior to ensure realistic forecasting of their impacts. Two driving behavior characteristics were collected using digital wave radar detectors, standstill distance and headway (CC0 and CC1 in Vissim, respectively). Three driving behavior models in Vissim were considered: car-following, lane-changing, and lateral behavior within the lane. The parameters of all three models were modified and validated using real-world trajectory data collected from a public test track. Data collection and calibration of CAV behavior in traffic were performed as part of the CoExist project (CoExist, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). The parameters were modified to simulate the ability of CAV to maintain a shorter headway, make instant reaction decisions, and achieve harmonized average speeds. In addition, the ability to communicate with other vehicles was simulated by enabling the waste collection truck to adjust its speed and gap acceptance preferences based on the types of surrounding vehicles. A sensitivity analysis was conducted to investigate the impact of varying traffic volume and operating speed on traffic delay, throughput, and carbon footprint. The selected factors can vary due to various reasons, such as traffic congestion at peak hours, extreme climatic conditions, governmental restrictions, and road accidents. A traffic volume of 70,000 vehicles and an operating speed of 45 km/h were utilized, with variations of 20,000 vehicles and \u0026plusmn;\u0026thinsp;5 km/h, respectively.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e\u003ch2\u003e3.8. Life Cycle Costing\u003c/h2\u003e\u003cp\u003eAn LCC was conducted to assess the economic performance of the examined waste collection systems. The analysis aimed to quantify the net present cost (NPC) of each system over a 15-year study period at a 5% discount rate. The capital expenditure (CAPEX) covered the procurement, installation, and periodic replacement of physical assets over their respective lifetimes, whereas the operational expenditure (OPEX) included fuel, labor, and maintenance costs derived from the simulated travel distances and service frequencies. Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e3\u003c/span\u003e breaks down the CAPEX and OPEX input parameters. The unit costs were obtained from international vendors at market prices to ensure realistic estimates.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eCost streams of the examined scenarios.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eParameter\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eCWC\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSWC\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eKWC\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBin (\u003cspan\u003e$\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e80\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e80\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e80\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLifespan of the bin (year)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e5\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBin smart sensor (\u003cspan\u003e$\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e70\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eProcessing and connectivity module (\u003cspan\u003e$\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e20\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLifespan of bin instrumentation (year)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMaintenance of equipped bin (% CAPEX)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e20\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eWaste collection vehicle (\u003cspan\u003e$\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e60,000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e60,000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e60,000\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eInstrumentation of autonomous vehicle (\u003cspan\u003e$\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e10,000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e10,000\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e10,000\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eOnboard vehicle weighing system (\u003cspan\u003e$\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3,000\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLifespan of onboard vehicle weighing system (year)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e5\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMaintenance of equipped vehicle (% CAPEX)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e20\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHourly rate of heavy vehicle driver (\u003cspan\u003e$\u003c/span\u003e/h)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e20\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e20\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e20\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e"},{"header":"4. Results and Discussion","content":"\u003cp\u003eThis section discusses the outcomes of the simulation analyses for data processing, waste generation forecasting, bin selection, route optimization, and autonomous operation phases, followed by a financial assessment. This discussion aims to compare the proposed knowledge-based autonomous smart collection system with typical sensor-based smart and traditional collection systems. This section investigates three scenarios: conventional waste collection, sensor-based waste collection, and knowledge-based waste collection.\u003c/p\u003e\n\u003cdiv id=\"Sec19\" class=\"Section2\"\u003e\n \u003ch2\u003e4.1. Data Processing\u003c/h2\u003e\n \u003cp\u003eFigure \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e shows the daily waste generation pattern in the study area during the analysis period. Waste generation fluctuated significantly daily, with higher and more variable quantities produced on average in 2019 compared to 2020. The daily generation in 2019 ranged from \u0026minus;\u0026thinsp;19.6% to 31.2% of the mean value of 4.7 tons, whereas it ranged between \u0026minus;\u0026thinsp;21.8% and 17.3% of the average value of 3.2 tons in 2020. The maximum generation in 2019 occurred in May and September, corresponding to the holy month of Ramadan and the beginning of the academic year, respectively. In contrast, the lowest amounts of waste were generated in January and August, during the study breaks. The waste generation in 2020 did not follow a pattern similar to that in 2019, as the lowest quantities were generated in May, whereas the highest waste production occurred in January, July, and August.\u003c/p\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e\n \u003ch2\u003e4.2. Waste Generation Prediction\u003c/h2\u003e\n \u003cp\u003ePredicting waste generation is one of the key differences between the proposed knowledge-based system, represented by the KWC scenario, and the sensor-based SWC scenario. The waste generation predictions, based on historical data acquired from onboard truck sensors, aim to replace the real-time data measured by bin-level or weight sensors. Table \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e lists the prediction results of the examined machine learning models to be used in the KWC scenario. The results indicated that both models efficiently predicted actual waste generation with R\u003csup\u003e2\u003c/sup\u003e values of 0.94 and 0.92 for XGBoost and the GLM, respectively. However, optimized XGBoost demonstrated a higher degree of accuracy than the GLM, showing a considerable enhancement in terms of the AE and RE by approximately 32% and 21%, respectively. Moreover, slight improvements of approximately 0.4% and 1.8% in the RMSE and SE, respectively, were observed.\u003c/p\u003e\n \u003cp\u003eTable 4: Comparative assessment of the prediction models in terms of various evaluation metrics.\u003c/p\u003e\n \u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" align=\"left\" width=\"99%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eModel\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRMSE\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAE\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRE\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSE\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eR\u003csup\u003e2\u003c/sup\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eGeneralized Linear Model (GLM)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e0.470\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e0.330\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e5.24%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e0.224\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e0.92\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 38px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eGradient Boosted Trees (XGBoost)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 15px;\"\u003e\n \u003cp\u003e0.468\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e0.224\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e4.16%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e0.220\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e0.94\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"6\" style=\"width: 100px;\"\u003e\n \u003cp\u003e\u003cem\u003eRoot Mean Squared Error (RMSE), Absolute Error (AE), Relative Error (RE), Squared Error (SE), Coefficient of Determination (R\u003csup\u003e2\u003c/sup\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003ctable id=\"Tab5\" border=\"1\"\u003e\u003c/table\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec21\" class=\"Section2\"\u003e\n \u003ch2\u003e4.3. Bin-Selection Algorithm\u003c/h2\u003e\n \u003cp\u003eThe heuristic bin-selection algorithm was applied in the SWC and KWC scenarios using measured and predicted waste data, respectively. Figure \u003cspan class=\"InternalRef\"\u003e7\u003c/span\u003e shows the daily variation in the number of serviced bins and the amount of waste collected in all scenarios. The number of bins and the amount of waste in the SWC and KWC scenarios varied significantly, with peak values occurring every 4\u0026ndash;5 days. In contrast, the CWC scenario exhibited minimal deviations in managed waste quantities, with a constant number of serviced bins. Furthermore, both the number of bins and the amount of waste followed the same pattern for the SWC and KWC scenarios.\u003c/p\u003e\n \u003cp\u003eTable \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e summarizes these variables in addition to the number of collection days. The CWC scenario had a fixed number of bins, approximately 396, collected every day, totaling 11,880 bins serviced in a month. In contrast, the SWC and KWC systems provided collection on only 25 and 27 days, respectively, as the daily generated waste did not exceed the bin capacity on certain days. Consequently, the total number of collected bins decreased significantly by 89% for both scenarios; the KWC scenario had only two fewer bins than the SWC scenario, despite the difference in minimum and maximum values. The SWC and KWC systems did not require servicing any bins in the first three days; hence, the waste accumulated until it reached full capacity. Moreover, despite the significant reduction in the number of bins selected in the SWC and KWC scenarios compared to the conventional system, the amount of collected waste was reduced by only approximately 9.97% and 10.61%, respectively. This highlights the potential improvements to conventional waste collection that can be achieved by replacing it with smart systems, which offer a similar level of service with significantly greater benefits.\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003ctable id=\"Tab6\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eStatistical bin-selection algorithm measures for the examined scenarios.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eVariable\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCWC\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSWC\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eKWC\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" rowspan=\"3\"\u003e\n \u003cp\u003e\u003cstrong\u003eNumber of Bins Serviced Daily\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eMinimum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e396\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e17\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eMaximum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e396\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e147\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e116\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eTotal\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e11,880\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1,288\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1,286\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" rowspan=\"3\"\u003e\n \u003cp\u003e\u003cstrong\u003eDaily Amount of Waste (ton)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eMinimum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.48\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.59\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.92\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eMaximum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8.50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6.65\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eTotal\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e83.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e74.76\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e74.22\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eNumber of Collection Days\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eTotal\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e25\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"5\"\u003e\n \u003cp\u003e\u003cem\u003eCWC: Conventional Waste Collection, SWC: Sensors-based Waste Collection, KWC: Knowledge-based Waste Collection\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec22\" class=\"Section2\"\u003e\n \u003ch2\u003e4.4. Route Optimization\u003c/h2\u003e\n \u003cp\u003eThe waste collection route was optimized for all scenarios, once for the CWC scenario and daily for the SWC and KWC scenarios. Figure \u003cspan class=\"InternalRef\"\u003e8\u003c/span\u003e displays the daily variations in the optimum route length for the SWC and KWC scenarios, whereas the CWC scenario had a constant daily travel distance because all bins were collected daily. The bin-selection algorithm showed that the system collected no waste on certain days in the SWC and KWC scenarios; consequently, the travel distance was zero. This resulted in an approximately 7 km decrease in the total SWC travel distance compared to the KWC scenario, as the latter required collection on extra days, although it had a slightly lower number of serviced bins.\u003c/p\u003e\n \u003cp\u003eFigure 9 shows the optimized CWC waste collection route, as well as the low- and high-service SWC and KWC days, along with several statistical measures of the travel distance for all scenarios. The travel distance of CWC was found to be 10.9 km, whereas it ranged between 1.5 and 8.4 km and 2.7 and 7.5 km for the SWC and KWC scenarios, respectively. In terms of the total travel distance, the SWC and KWC scenarios significantly outperformed the CWC scenario by 63.1% and 60.9%, respectively.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec23\" class=\"Section2\"\u003e\n \u003ch2\u003e4.5. Autonomous Navigation\u003c/h2\u003e\n \u003cp\u003eThis study simulated autonomous navigation for all scenarios to investigate its individual effect on conventional and smart waste collection systems. Table \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003e illustrates the application of CAVs against human-driven vehicles in all assessed scenarios. The CWC scenario had the highest carbon footprint, whereas the SWC and KWC scenarios demonstrated significantly better performance, minimizing total emissions by 49% and 48% for human-driven vehicles and 49% and 50% for CAVs, respectively, due to their considerably lower total travel distance and number of collection days. In addition, the total emissions were decreased by CAVs in all scenarios, with the KWC scenario achieving the highest reduction of 29%. Similarly, the CWC scenario resulted in the longest travel time, while the SWC and KWC scenarios significantly reduced the total travel time by approximately 87% and 85%, respectively, for both regular and autonomous vehicles. In contrast, regular and autonomous vehicles demonstrated a similar performance, as there was no traffic congestion. Moreover, the SWC and KWC scenarios experienced no or low traffic delays, as several days did not require collections, and queued vehicles did not obstruct the waste collection trucks due to relatively shorter routes compared to the CWC scenario, which had the longest traffic delays. However, the maximum delay of the CWC scenario was 3.24 seconds using CAVs, outperforming the SWC scenario by 67%. The implementation of CAVs significantly improved the average traffic delay by 83%, 68%, and 73% for the CWC, SWC, and KWC scenarios, respectively. Furthermore, the CWC scenario resulted in the longest queue length, whereas the SWC and KWC scenarios achieved reductions of 20% and 11%, respectively, using human-driven collection vehicles. These reductions approximately doubled when using CAVs, with values of approximately 42% and 21%, respectively. Additionally, autonomous collection trucks enhanced the average queue length by 46%, 61%, and 52% in the CWC, SWC, and KWC scenarios, respectively.\u003c/p\u003e\n \u003cp\u003eTable 6: Traffic simulation results for regular and autonomous vehicles.\u003c/p\u003e\n \u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eVariable\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eVehicle Type\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eCWC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eSWC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eKWC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"6\"\u003e\n \u003cp\u003e\u003cstrong\u003eTravel Time (h)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\"\u003e\n \u003cp\u003e\u003cstrong\u003eRegular\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eMinimum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e6.58\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.48\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.51\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eMaximum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e7.04\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.68\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eTotal\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e173.91\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e22.94\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e26.32\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\"\u003e\n \u003cp\u003e\u003cstrong\u003eAutonomous\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eMinimum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e6.53\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.46\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.52\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eMaximum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e6.84\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.91\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.58\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eTotal\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e172.75\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e22.39\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e25.88\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"6\"\u003e\n \u003cp\u003e\u003cstrong\u003eQueue Length (m)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\"\u003e\n \u003cp\u003e\u003cstrong\u003eRegular\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eMinimum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e26.43\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e28.14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e24.11\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eMaximum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e42.85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e40.60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e43.16\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eAverage\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e34.87\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e28.06\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e31.07\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\"\u003e\n \u003cp\u003e\u003cstrong\u003eAutonomous\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eMinimum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e11.32\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e11.18\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e10.35\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eMaximum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e23.57\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e18.28\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e18.95\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eAverage\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e18.84\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e10.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e14.95\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"6\"\u003e\n \u003cp\u003e\u003cstrong\u003eTraffic Delay (s)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\"\u003e\n \u003cp\u003e\u003cstrong\u003eRegular\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eMinimum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e17.42\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e2.87\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e2.79\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eMaximum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e19.93\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e11.33\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e11.38\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eAverage\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e18.63\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e5.40\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e6.82\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\"\u003e\n \u003cp\u003e\u003cstrong\u003eAutonomous\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eMinimum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e2.97\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.84\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.68\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eMaximum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e3.24\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e5.41\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e2.35\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eAverage\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e3.12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.73\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.86\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"6\"\u003e\n \u003cp\u003e\u003cstrong\u003eCarbon Footprint (CO\u003csub\u003e2\u003c/sub\u003e-eq)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\"\u003e\n \u003cp\u003e\u003cstrong\u003eRegular\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eMinimum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.94\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.44\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.33\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eMaximum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.92\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.80\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eTotal\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e28.22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e14.29\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e14.69\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\"\u003e\n \u003cp\u003e\u003cstrong\u003eAutonomous\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eMinimum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.68\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.23\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eMaximum\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e1.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.68\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e0.62\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eTotal\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e21.16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e10.69\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e10.50\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"6\"\u003e\n \u003cp\u003e\u003cem\u003eCWC: Conventional Waste Collection, SWC: Sensors-based Waste Collection, KWC: Knowledge-based Waste Collection\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003cp\u003eFigure 10 illustrates the impact of varying the traffic volume and operating speed on traffic delay, throughput, and carbon footprint in all scenarios using both regular and autonomous waste collection vehicles. The model reveals significant changes when traffic volume is increased, particularly in terms of traffic delay and carbon footprint, compared to when it is reduced. For instance, the percentage change in the traffic delay and carbon footprint when increasing the traffic volume for all scenarios ranged from -260% to -100% and from -200% to -70%, respectively, whereas the maximum change realized by decreasing the traffic volume reached approximately 50% and 30%, respectively. Overall, the impact of CAVs was more significant at higher traffic volumes and lower operating speeds, as the benefits of CAVs cannot be fully realized in congested traffic.\u003c/p\u003e\n \u003cp\u003eFigure \u003cspan class=\"InternalRef\"\u003e11\u003c/span\u003e shows a heatmap of the traffic volume, in vehicle per hour (vph), for the most utilized streets in the study area for the CWC scenario, as well as the low- and high-service days of the SWC and KWC scenarios. The CWC scenario had the most congested roads, as all bins were collected, followed by the SWC and KWC scenarios, which required fewer bins to be collected.\u003c/p\u003e\n \u003ctable id=\"Tab7\" border=\"1\"\u003e\u003c/table\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec24\" class=\"Section2\"\u003e\n \u003ch2\u003e4.6. Economic Feasibility\u003c/h2\u003e\n \u003cp\u003eFigure \u003cspan class=\"InternalRef\"\u003e12\u003c/span\u003e presents the annual NPC for the scenarios examined in the study area over a 15-year evaluation period. The investment costs included capital expenditures for waste collection vehicles, bins, and their related instrumentation. The periodic cost spikes every five years correspond to the scheduled replacement of bins and sensors in accordance with their expected service life. The CWC scenario had an NPC of approximately \u003cspan\u003e$\u003c/span\u003e580k, mainly due to high fuel consumption and labor intensity from daily fixed-route operations. The SWC system incurred the highest NPC, at approximately \u003cspan\u003e$\u003c/span\u003e609k; although optimized routing reduced fuel usage, the savings were insufficient to offset the significant expenses of sensors, connectivity modules, and their recurring maintenance, resulting in a total cost 5% higher than that of CWC. On the other hand, despite using a more costly sensor, the KWC system achieved a significantly lower NPC of \u003cspan\u003e$\u003c/span\u003e196k, representing a 65% reduction relative to CWC and SWC. This substantial reduction is primarily attributed to the elimination of hundreds of bin-mounted sensors and their associated installation and maintenance costs. By centralizing data collection through a single onboard weighing system and leveraging predictive, cloud-based analytics, KWC minimized hardware dependence while maintaining high operational efficiency. When CAV configuration was introduced, the cost of vehicle electrification was incorporated, while fuel and labor costs were reduced due to automation and shorter collection times. Despite these operational savings, CAV scenarios exhibited slightly higher NPCs compared to their conventional counterparts: \u003cspan\u003e$\u003c/span\u003e586k, \u003cspan\u003e$\u003c/span\u003e627k, and \u003cspan\u003e$\u003c/span\u003e223k for CWC-CAV, SWC-CAV, and KWC-CAV, respectively. The increase is mainly attributed to the added costs of instrumentation for autonomous vehicles. These findings underscore the significant economic and operational benefits of data-driven, cloud-based architectures over hardware-intensive systems.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec25\" class=\"Section2\"\u003e\n \u003ch2\u003e4.7. Overall Assessment\u003c/h2\u003e\n \u003cp\u003eFigure 13 compares the outcomes of the bin-selection algorithm, route optimization, and autonomous operation analyses for the smart waste collection scenarios as a percentage reduction from the baseline scenario (CWC). Similar results were obtained using both regular and autonomous vehicles in the bin-selection algorithm, and route optimization analyses were assessed in terms of autonomous operation through metrics such as travel time, queue length, traffic delay, carbon footprint, and economic feasibility. The SWC and KWC systems reduced the total number of collected bins by approximately 89%, as the bin-selection algorithm selected only full bins. This led to avoiding collection on five and three days in the SWC and KWC scenarios, respectively, as no bins required servicing; thus, the amount of collected waste was approximately 10% lower in both scenarios. In addition, the total travel distance decreased by 63% and 61% in the SWC and KWC scenarios, respectively, as the daily number of selected bins was consistently less than that in the CWC scenario. Moreover, the total travel time of both regular and autonomous vehicles decreased by approximately 87% and 85% for the SWC and KWC scenarios, respectively. Using regular vehicles in the SWC and KWC scenarios increased the carbon footprint by 49% and 48%, respectively, compared to the conventional scenario, whereas CWC automation led to a 25% improvement. However, the latter contributed to an 83% reduction in traffic delay and a 46% reduction in queue length, respectively. From an economic perspective, KWC achieved substantially lower NPC by 63%, whereas SWC incurred 7% higher costs due to the additional hardware and maintenance expenses. In general, the automation slightly increased the NPC due to the added vehicle instrumentation costs, which were not fully offset by reduced operation expenses. Overall, the best performance in terms of traffic delay, queue length, and carbon footprint was realized by utilizing autonomous vehicles in the SWC and KWC scenarios.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec26\" class=\"Section2\"\u003e\n \u003ch2\u003e4.8. Potential Applications and Challenges\u003c/h2\u003e\n \u003cp\u003eBeyond the waste management sector, this framework can be adopted in a broad range of applications, as supply forecasting, resource selection, and route optimization are not unique to waste logistics. Similar architectures could be tailored to other industries where demand is variable and spatially distributed, such as parcel delivery, food distribution, or recycling of electronic waste. Even outside urban services, the same principles could be applied to industrial supply chains; for instance, optimizing the collection of agricultural produce, medical supplies, or construction debris. Embedding predictive analytics and autonomous routing into such domains highlights the versatility of the framework and its potential contribution to smart city systems and industrial sustainability goals.\u003c/p\u003e\n \u003cp\u003eNevertheless, from a practical implementation perspective, several challenges may arise, particularly in scaling the system from simulation to large-scale deployment. For example, the maintenance of onboard sensors, the integration of autonomous vehicles into existing traffic regulations, and the interoperability between municipal platforms are all issues that operators must address before widespread adoption. Real-world pilot projects would be required to evaluate robustness under operational uncertainties, such as irregular waste generation, extreme weather conditions, or hardware malfunctions. Nevertheless, the proposed system reduces reliance on bin sensors, which lowers infrastructure costs and improves scalability compared to traditional hardware‐intensive approaches.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"5. Conclusion","content":"\u003cp\u003eA novel autonomous cloud-based smart waste collection system has been proposed to replace conventional systems, reducing capital investments, operating costs, traffic congestion, and air pollution emissions. The system integrates various components: 1) AI models predicting the daily waste generation for each bin, 2) a bin-selection algorithm prioritizing bins requiring collection, 3) optimization techniques identifying the best collection route, and 4) autonomous navigation providing driverless waste collection. The proposed framework was applied through computer simulations of a residential area in Abu Dhabi, UAE, which contained 396 bins. For this area, 2-year historical records of daily generated recyclable waste per bin were obtained. Three scenarios were compared: CWC, SWC, and KWC. CWC represented the conventional system, in which all bins were collected daily through the same route. SWC represented a typical sensor-based smart system, and KWC demonstrated the proposed system.\u003c/p\u003e\u003cp\u003eThe GLM and XGBoost models were tested to predict daily waste quantities, where the latter outperformed the former with an absolute error of 0.22. The KWC and SWC scenarios utilized waste quantities forecasted by the XGBoost model and the actual quantities generated, respectively, to select the bins requiring collection using a bin-selection algorithm. Both the SWC and KWC achieved significant reductions in the total number of serviced bins, selecting 89% fewer bins than the CWC. Next, the collection routes for each scenario were optimized using GIS methods to determine the shortest travel path. The total CWC travel distance improved by 63% and 61% in the SWC and KWC scenarios, respectively. The operation of autonomous vehicles was analyzed and compared to regular driving in all examined scenarios. The total delay improved significantly by 91% and 90% in the automated SWC and KWC scenarios, respectively, whereas the CWC scenario had the longest traffic delay and travel time. Moreover, an LCC analysis was conducted to assess the economic performance of the systems over a 15-year period. The results revealed that SWC incurred the highest costs due to extensive sensor installation and maintenance, while KWC achieved a 65% cost reduction relative to SWC by replacing hardware-intensive components with a cloud-based system. Overall, the technical performance of the KWC and SWC systems was close; nevertheless, the KWC system is distinctly advantageous over the SWC system because it eliminates the capital and operating costs of the bin sensory data acquisition and communication system, offering a more cost-effective, eco-friendly, and less complex smart system to establish and maintain.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eFunding\u003c/h2\u003e\u003cp\u003eThis research received no specific grant from any funding agency in the public, private, or not-for-profit sectors.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eM.A: conceptualization, formal analysis, methodology, data curation, supervision, validation, and writing\u0026mdash;review and editing, and M.H: formal analysis, methodology, software, writing\u0026mdash;original draft.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe data supporting the findings of this study were provided by Abu Dhabi Waste Management Center (Tadweer), but restrictions apply to the availability of the data. The daily waste generation records, used under license from Tadweer for the purpose of this study, are not publicly available.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAbbasi, M., Rastgoo, M. N. \u0026amp; Nakisa, B. Monthly and seasonal modeling of municipal waste generation using radial basis function neural network. \u003cem\u003eEnviron. Progress Sustainable Energy\u003c/em\u003e. \u003cb\u003e38\u003c/b\u003e (3). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/ep.13033\u003c/span\u003e\u003cspan address=\"10.1002/ep.13033\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2019).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAbdallah, M., Adghim, M., Maraqa, M. \u0026amp; Aldahab, E. Simulation and optimization of dynamic waste collection routes. \u003cem\u003eWaste Manage. Res.\u003c/em\u003e \u003cb\u003e37\u003c/b\u003e (8), 793\u0026ndash;802. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/0734242X19833152\u003c/span\u003e\u003cspan address=\"10.1177/0734242X19833152\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2019).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAdusei, K. K., Ng, K. T. W., Mahmud, T. S., Karimi, N. \u0026amp; Lakhan, C. Exploring the use of astronomical seasons in municipal solid waste disposal rates modeling. \u003cem\u003eSustainable Cities and Society\u003c/em\u003e, \u003cem\u003e86\u003c/em\u003e. (2022). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.scs.2022.104115\u003c/span\u003e\u003cspan address=\"10.1016/j.scs.2022.104115\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAlfeo, A. L. et al. Urban swarms: A new approach for autonomous waste management. \u003cem\u003eProceedings - IEEE International Conference on Robotics and Automation\u003c/em\u003e, \u003cem\u003e2019-May\u003c/em\u003e, 4233\u0026ndash;4240. (2019). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/ICRA.2019.8794020\u003c/span\u003e\u003cspan address=\"10.1109/ICRA.2019.8794020\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAlsobky, A., Ahmed, M., Agroudy, A., Araby, E. \u0026amp; S., \u0026amp;, K A smart framework for municipal solid waste collection management: A case study in Greater Cairo Region. \u003cem\u003eAin Shams Eng. J.\u003c/em\u003e \u003cb\u003e14\u003c/b\u003e (6). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.asej.2023.102183\u003c/span\u003e\u003cspan address=\"10.1016/j.asej.2023.102183\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAmal, L., Son, L. H. \u0026amp; Chabchoub, H. SGA: spatial GIS-based genetic algorithm for route optimization of municipal solid waste collection. \u003cem\u003eEnviron. Sci. Pollut. Res.\u003c/em\u003e \u003cb\u003e25\u003c/b\u003e (27), 27569\u0026ndash;27582. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11356-018-2826-0\u003c/span\u003e\u003cspan address=\"10.1007/s11356-018-2826-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2018).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAraiza-Aguilar, J. A., Rojas-Valencia, M. N. \u0026amp; Aguilar-Vera, R. A. Forecast generation model of municipal solid waste using multiple linear regression. \u003cem\u003eGlobal J. Environ. Sci. Manage.\u003c/em\u003e \u003cb\u003e6\u003c/b\u003e (1), 1\u0026ndash;14. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.22034/gjesm.2020.01.01\u003c/span\u003e\u003cspan address=\"10.22034/gjesm.2020.01.01\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2020).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBelhiah, M., El Aboudi, M. \u0026amp; Ziti, S. Optimising unplanned waste collection: An IoT-enabled system for smart cities, a case study in Tangier, Morocco. \u003cem\u003eIET Smart Cities\u003c/em\u003e. \u003cb\u003e6\u003c/b\u003e (1), 27\u0026ndash;40. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1049/smc2.12069\u003c/span\u003e\u003cspan address=\"10.1049/smc2.12069\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBrouwer, Y., Barbosa-P\u0026oacute;voa, A. P., Antunes, A. P., Rodrigues, P. \u0026amp; Ramos, T. Comparison of different waste bin monitoring approaches: An exploratory study. \u003cem\u003eWaste Manage. Res.\u003c/em\u003e \u003cb\u003e41\u003c/b\u003e (10), 1570\u0026ndash;1583. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/0734242X231160691\u003c/span\u003e\u003cspan address=\"10.1177/0734242X231160691\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCatania, V. \u0026amp; Ventura, D. An approch for monitoring and smart planning of urban solid waste management using smart-M3 platform. \u003cem\u003eConference of Open Innovation Association, FRUCT\u003c/em\u003e, 24\u0026ndash;31. (2014). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/FRUCT.2014.6872422\u003c/span\u003e\u003cspan address=\"10.1109/FRUCT.2014.6872422\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCoExist. \u003cem\u003ePTV Vissim Tool\u003c/em\u003e. (2019).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFan, L. et al. Introducing an evolutionary-decomposition model for prediction of municipal solid waste flow: application of intrinsic time-scale decomposition algorithm. \u003cem\u003eEng. Appl. Comput. Fluid Mech.\u003c/em\u003e \u003cb\u003e15\u003c/b\u003e (1), 1159\u0026ndash;1175. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/19942060.2021.1945496\u003c/span\u003e\u003cspan address=\"10.1080/19942060.2021.1945496\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGhanbari, F., Kamalan, H. \u0026amp; Sarraf, A. An evolutionary machine learning approach for municipal solid waste generation estimation utilizing socioeconomic components. \u003cem\u003eArab. J. Geosci.\u003c/em\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s12517-020-06348-w/Published\u003c/span\u003e\u003cspan address=\"10.1007/s12517-020-06348-w/Published\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGreco, S. A. M., Rossit, D. G., Frutos, M. \u0026amp; Cavallin, A. Optimization of waste collection through the sequencing of micro-routes and transfer station convenience analysis: an Argentinian case study. \u003cem\u003eWaste Management \u0026amp; Research\u003c/em\u003e. (2023). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://gobiernoabierto.bahia.gob.ar/ambientales/residuos-domiciliarios/\u003c/span\u003e\u003cspan address=\"https://gobiernoabierto.bahia.gob.ar/ambientales/residuos-domiciliarios/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGuha, B., Momtaz, Z., Kafy, A., Al \u0026amp; Rahaman, Z. A. Estimating solid waste generation and suitability analysis of landfill sites using regression, geospatial, and remote sensing techniques in Rangpur, Bangladesh. \u003cem\u003eEnviron. Monit. Assess.\u003c/em\u003e \u003cb\u003e195\u003c/b\u003e (1). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10661-022-10695-4\u003c/span\u003e\u003cspan address=\"10.1007/s10661-022-10695-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGutierrez, J. M., Jensen, M., Henius, M. \u0026amp; Riaz, T. Smart Waste Collection System Based on Location Intelligence. \u003cem\u003eProcedia Comput. Sci.\u003c/em\u003e \u003cb\u003e61\u003c/b\u003e, 120\u0026ndash;127. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.procs.2015.09.170\u003c/span\u003e\u003cspan address=\"10.1016/j.procs.2015.09.170\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2015).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHoang, M. G., Fujiwara, T., Pham Phu, S. T. \u0026amp; Nguyen Thi, K. T. Predicting waste generation using Bayesian model averaging. \u003cem\u003eGlobal J. Environ. Sci. Manage.\u003c/em\u003e \u003cb\u003e3\u003c/b\u003e (4), 385\u0026ndash;402. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.22034/gjesm.2017.03.04.005\u003c/span\u003e\u003cspan address=\"10.22034/gjesm.2017.03.04.005\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2017).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHu, S., An, L. \u0026amp; Shen, L. A multi-objective modeling and optimization approach to municipal solid waste collection for classified treatment in China towards sustainable development. \u003cem\u003eSustainable Cities and Society\u003c/em\u003e, \u003cem\u003e98\u003c/em\u003e. (2023). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.scs.2023.104846\u003c/span\u003e\u003cspan address=\"10.1016/j.scs.2023.104846\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHussain, D. I., Elomri, D. A., Kerbache, D. L., Omri, D. A. \u0026amp; El Smart city solutions: Comparative analysis of waste management models in IoT-enabled environments using multiagent simulation. \u003cem\u003eSustainable Cities and Society\u003c/em\u003e, \u003cem\u003e103\u003c/em\u003e. (2024). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.scs.2024.105247\u003c/span\u003e\u003cspan address=\"10.1016/j.scs.2024.105247\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJassim, M. S., Coskuner, G. \u0026amp; Zontul, M. Comparative performance analysis of support vector regression and artificial neural network for prediction of municipal solid waste generation. \u003cem\u003eWaste Manage. Res.\u003c/em\u003e \u003cb\u003e40\u003c/b\u003e (2), 195\u0026ndash;204. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/0734242X211008526\u003c/span\u003e\u003cspan address=\"10.1177/0734242X211008526\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJorge, D., Pais Antunes, A., Rodrigues Pereira Ramos, T. \u0026amp; Barbosa-P\u0026oacute;voa, A. P. A hybrid metaheuristic for smart waste collection problems with workload concerns. \u003cem\u003eComputers and Operations Research\u003c/em\u003e, \u003cem\u003e137\u003c/em\u003e. (2022). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.cor.2021.105518\u003c/span\u003e\u003cspan address=\"10.1016/j.cor.2021.105518\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJusto, V. B., Gupta, A., Umland, T. F. \u0026amp; G\u0026ouml;hlich, D. Minimum Energy Utilization Strategy for Fleet of Autonomous Robots in Urban Waste Management. \u003cem\u003eRobotics\u003c/em\u003e \u003cb\u003e12\u003c/b\u003e (6). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/robotics12060159\u003c/span\u003e\u003cspan address=\"10.3390/robotics12060159\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJuwariyah, T., Krisnawati, L. \u0026amp; Sulasminingsih, S. Design of IoT-Based Smart Bins Integrated Monitoring System Using Blynk. \u003cem\u003eIOP Conference Series: Materials Science and Engineering\u003c/em\u003e, \u003cem\u003e1125\u003c/em\u003e(1), 012078. (2021). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1088/1757-899x/1125/1/012078\u003c/span\u003e\u003cspan address=\"10.1088/1757-899x/1125/1/012078\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKaratagi, A., Babu Peter, R., Shubham, R., Biradar, R. C. \u0026amp; Devanagavi, G. D. Integrating Electric Vehicle (EV) in Waste Management for Garbage Collection using LoRa Technology. \u003cem\u003eProceedings \u0026ndash;\u0026thinsp;2022 6th International Conference on Intelligent Computing and Control\u003c/em\u003e Systems, \u003cem\u003eICICCS 2022\u003c/em\u003e, 121\u0026ndash;128. (2022). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/ICICCS53718.2022.9788416\u003c/span\u003e\u003cspan address=\"10.1109/ICICCS53718.2022.9788416\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKaya, S. A hybrid firefly and particle swarm optimization algorithm with local search for the problem of municipal solid waste collection: a real-life example. \u003cem\u003eNeural Comput. Appl.\u003c/em\u003e \u003cb\u003e35\u003c/b\u003e (9), 7107\u0026ndash;7124. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s00521-022-08173-6\u003c/span\u003e\u003cspan address=\"10.1007/s00521-022-08173-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKontokosta, C. E., Hong, B., Johnson, N. E. \u0026amp; Starobin, D. Using machine learning and small area estimation to predict building-level municipal solid waste generation in cities. \u003cem\u003eComput. Environ. Urban Syst.\u003c/em\u003e \u003cb\u003e70\u003c/b\u003e, 151\u0026ndash;162. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.compenvurbsys.2018.03.004\u003c/span\u003e\u003cspan address=\"10.1016/j.compenvurbsys.2018.03.004\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2018).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLatif, S. D., Hazrin, N. A. B., Younes, M. K., Ahmed, A. N. \u0026amp; Elshafie, A. Evaluating different machine learning models for predicting municipal solid waste generation: a case study of Malaysia. \u003cem\u003eEnviron. Dev. Sustain.\u003c/em\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10668-023-03882-x\u003c/span\u003e\u003cspan address=\"10.1007/s10668-023-03882-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLiang, G. et al. Predicting municipal solid waste using a coupled artificial neural network with archimedes optimisation algorithm and socioeconomic components. \u003cem\u003eJournal of Cleaner Production\u003c/em\u003e, \u003cem\u003e315\u003c/em\u003e. (2021). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jclepro.2021.128039\u003c/span\u003e\u003cspan address=\"10.1016/j.jclepro.2021.128039\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLu, W., Huo, W., Gulina, H. \u0026amp; Pan, C. Development of machine learning multi-city model for municipal solid waste generation prediction. \u003cem\u003eFront. Environ. Sci. Eng.\u003c/em\u003e \u003cb\u003e16\u003c/b\u003e (9). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11783-022-1551-6\u003c/span\u003e\u003cspan address=\"10.1007/s11783-022-1551-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLuo, K., Zhao, W. \u0026amp; Zhang, R. A multi-day waste collection and transportation problem with selective collection and split delivery. \u003cem\u003eAppl. Math. Model.\u003c/em\u003e \u003cb\u003e126\u003c/b\u003e, 753\u0026ndash;771. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.apm.2023.11.009\u003c/span\u003e\u003cspan address=\"10.1016/j.apm.2023.11.009\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMah\u0026eacute;o, A., Rossit, D. G. \u0026amp; Kilby, P. Solving the integrated bin allocation and collection routing problem for municipal solid waste: a Benders decomposition approach. \u003cem\u003eAnn. Oper. Res.\u003c/em\u003e \u003cb\u003e322\u003c/b\u003e (1), 441\u0026ndash;465. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10479-022-04918-7\u003c/span\u003e\u003cspan address=\"10.1007/s10479-022-04918-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMamun, M. A., Al, Hannan, M. A., Hussain, A. \u0026amp; Basri, H. Theoretical model and implementation of a real time intelligent bin status monitoring system using rule based decision algorithms. \u003cem\u003eExpert Syst. Appl.\u003c/em\u003e \u003cb\u003e48\u003c/b\u003e, 76\u0026ndash;88. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.eswa.2015.11.025\u003c/span\u003e\u003cspan address=\"10.1016/j.eswa.2015.11.025\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2016).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMartikkala, A., Mayanti, B., Helo, P., Lobov, A. \u0026amp; Ituarte, I. F. Smart textile waste collection system \u0026ndash; Dynamic route optimization with IoT. \u003cem\u003eJournal of Environmental Management\u003c/em\u003e, \u003cem\u003e335\u003c/em\u003e. (2023). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jenvman.2023.117548\u003c/span\u003e\u003cspan address=\"10.1016/j.jenvman.2023.117548\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMohammadi, M. et al. A dynamic approach for the multi-compartment vehicle routing problem in waste management. \u003cem\u003eRenewable and Sustainable Energy Reviews\u003c/em\u003e, \u003cem\u003e184\u003c/em\u003e. (2023). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.rser.2023.113526\u003c/span\u003e\u003cspan address=\"10.1016/j.rser.2023.113526\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMudannayake, O., Rathnayake, D., Herath, J. D., Fernando, D. K. \u0026amp; Fernando, M. Exploring Machine Learning and Deep Learning Approaches for Multi-Step Forecasting in Municipal Solid Waste Generation. \u003cem\u003eIEEE Access.\u003c/em\u003e \u003cb\u003e10\u003c/b\u003e, 122570\u0026ndash;122585. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/ACCESS.2022.3221941\u003c/span\u003e\u003cspan address=\"10.1109/ACCESS.2022.3221941\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNamoun, A. et al. An Ensemble Learning Based Classification Approach for the Prediction of Household Solid Waste Generation. \u003cem\u003eSensors\u003c/em\u003e \u003cb\u003e22\u003c/b\u003e (9). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/s22093506\u003c/span\u003e\u003cspan address=\"10.3390/s22093506\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNoufal, M., Yuanyuan, L., Maalla, Z. \u0026amp; Adipah, S. Determinants of household solid waste generation and composition in Homs city, Syria. \u003cem\u003eJournal of Environmental and Public Health\u003c/em\u003e, \u003cem\u003e2020\u003c/em\u003e. (2020). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1155/2020/7460356\u003c/span\u003e\u003cspan address=\"10.1155/2020/7460356\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRambandara, R. D. S. S., Prabodanie, R. A. R., Karunarathne, E. A. C. P. \u0026amp; Rajapaksha, R. D. D. Improving the Efficiency of Urban Waste Collection Using Optimization: a Case Study. \u003cem\u003eProcess. Integr. Optim. Sustain.\u003c/em\u003e \u003cb\u003e6\u003c/b\u003e (3), 809\u0026ndash;818. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s41660-022-00232-8\u003c/span\u003e\u003cspan address=\"10.1007/s41660-022-00232-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRamos, T. R. P., de Morais, C. S. \u0026amp; Barbosa-P\u0026oacute;voa, A. P. The smart waste collection routing problem: Alternative operational management approaches. \u003cem\u003eExpert Syst. Appl.\u003c/em\u003e \u003cb\u003e103\u003c/b\u003e, 146\u0026ndash;158. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.eswa.2018.03.001\u003c/span\u003e\u003cspan address=\"10.1016/j.eswa.2018.03.001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2018).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSAE. \u003cem\u003eTaxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles\u003c/em\u003e. (2018).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSidhu, N., Pons-Buttazzo, A., Mu\u0026ntilde;oz, A. \u0026amp; Terroso-Saenz, F. A collaborative application for assisting the management of household plastic waste through smart bins: A case of study in the Philippines. \u003cem\u003eSensors\u003c/em\u003e \u003cb\u003e21\u003c/b\u003e (13). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/s21134534\u003c/span\u003e\u003cspan address=\"10.3390/s21134534\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSong, J. et al. Simulated annealing based hybrid forecast for improving daily municipal solid waste generation prediction. \u003cem\u003eScientific World Journal\u003c/em\u003e, \u003cem\u003e2014\u003c/em\u003e. (2014). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1155/2014/834357\u003c/span\u003e\u003cspan address=\"10.1155/2014/834357\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSrivastava, A. \u0026amp; Jha, P. K. A multi-model forecasting approach for solid waste generation by integrating demographic and socioeconomic factors: a case study of Prayagraj, India. \u003cem\u003eEnviron. Monit. Assess.\u003c/em\u003e \u003cb\u003e195\u003c/b\u003e (6). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10661-023-11338-y\u003c/span\u003e\u003cspan address=\"10.1007/s10661-023-11338-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSun Yin, H. \u0026amp; Vatrapu, R. A first estimation of the proportion of cybercriminal entities in the bitcoin ecosystem using supervised machine learning. \u003cem\u003eProceedings \u0026ndash;\u0026thinsp;2017 IEEE International Conference on Big Data, Big Data 2017\u003c/em\u003e, \u003cem\u003e2018-Janua\u003c/em\u003e, 3690\u0026ndash;3699. (2017). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/BigData.2017.8258365\u003c/span\u003e\u003cspan address=\"10.1109/BigData.2017.8258365\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang, S., Zhang, J., Zhao, Z. \u0026amp; Xin, C. Robust Optimization of Municipal Solid Waste Collection and Transportation with Uncertain Waste Output: A Case Study. \u003cem\u003eJ. Syst. Sci. Syst. Eng.\u003c/em\u003e \u003cb\u003e31\u003c/b\u003e (2), 204\u0026ndash;225. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11518-021-5510-8\u003c/span\u003e\u003cspan address=\"10.1007/s11518-021-5510-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZheng, B. \u0026amp; Agresti, A. \u003cem\u003eSummarizing the predictive power of a generalized linear model\u003c/em\u003e. \u003cem\u003eDecember 1998\u003c/em\u003e, 1771\u0026ndash;1781. (2000).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8012685/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8012685/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eSubstantial efforts have been directed to modernize and decarbonize the waste collection industry. Smart sensor-based waste collection (SWC) systems were developed to optimize collection routes based on the actual waste levels in bins, resulting in reduced service frequency, fuel consumption, and air pollution. To date, there has not been a fully functional commercial SWC system due to the complexity and limited practicality of such a hardware-intensive design. Alternatively, this research presents an innovative cloud-based approach in which historical waste generation data, acquired from onboard-truck sensors, replaces data from bin sensors. The proposed knowledge-based waste collection (KWC) system incorporates machine learning for waste forecasting, expert bin selection, route optimization, and autonomous navigation. The system was simulated in an actual residential district to collect recyclables over three scenarios: conventional, SWC, and KWC. Historical data were used to train a generalized linear model (GLM) and a gradient-boosted tree (XGBoost) model to predict daily waste generation per bin. The heuristic bin-selection algorithm selected the bins to be served based on the actual and forecasted waste quantities in SWC and KWC, respectively. XGBoost achieved higher prediction accuracy than GLM with a 4.2% relative error. SWC and KWC significantly reduced the travel distance by 63.1% and 60.9%, respectively, whereas the number of collected bins decreased by 89% for both scenarios. The number of collection days decreased by 5 and 3 days per month in SWC and KWC, respectively. The implementation of connected and autonomous vehicles (CAVs) significantly improved the system, decreasing the total delay by 91% and 90% in SWC and KWC, respectively. Moreover, a life cycle costing analysis revealed that, compared to conventional collection, the reduced travel expenses of SWC were insufficient to offset the cost of bin sensors, whereas KWC achieved a 63% cost reduction by replacing hardware-intensive components with a cloud-based system. Overall, this research has demonstrated that KWC systems can potentially outperform hardware-intensive SWC systems, given the substantial economic and operational benefits associated with such a cloud-based approach.\u003c/p\u003e","manuscriptTitle":"Development and Simulation of an Innovative Autonomous Knowledge-Based Smart Waste Collection System","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-17 09:19:12","doi":"10.21203/rs.3.rs-8012685/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-12-04T08:27:24+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-03T16:35:33+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-03T15:53:22+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-01T05:03:34+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"93305608039875172499866245958049013261","date":"2025-11-28T03:44:24+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"67099493925091036543291060402963708062","date":"2025-11-24T19:26:35+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"182112054685212489698779290838541654601","date":"2025-11-23T06:57:52+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"89538975947250112428172511732320453164","date":"2025-11-22T22:49:21+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"321210520721517659921050641464281351846","date":"2025-11-06T08:09:27+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"89821070149464972582290801821271450591","date":"2025-11-06T08:02:25+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-11-06T07:56:41+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-11-06T06:54:36+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-11-04T08:33:04+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-11-04T08:31:43+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-11-02T18:20:58+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"de726772-86ff-48f7-978f-2948b666b6aa","owner":[],"postedDate":"November 17th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":58058652,"name":"Physical sciences/Energy science and technology"},{"id":58058653,"name":"Physical sciences/Engineering"},{"id":58058654,"name":"Earth and environmental sciences/Environmental sciences"},{"id":58058655,"name":"Physical sciences/Mathematics and computing"}],"tags":[],"updatedAt":"2026-04-27T16:08:51+00:00","versionOfRecord":{"articleIdentity":"rs-8012685","link":"https://doi.org/10.1038/s41598-026-48792-w","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2026-04-26 16:00:16","publishedOnDateReadable":"April 26th, 2026"},"versionCreatedAt":"2025-11-17 09:19:12","video":"","vorDoi":"10.1038/s41598-026-48792-w","vorDoiUrl":"https://doi.org/10.1038/s41598-026-48792-w","workflowStages":[]},"version":"v1","identity":"rs-8012685","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8012685","identity":"rs-8012685","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}