Hierarchical Environmental Semantic Reconstruction for Stable Mobility of Visually Impaired Pedestrians on Urban Streets in Shanghai

Hierarchical Environmental Semantic Reconstruction for Stable Mobility of Visually Impaired Pedestrians on Urban Streets in Shanghai | 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 Hierarchical Environmental Semantic Reconstruction for Stable Mobility of Visually Impaired Pedestrians on Urban Streets in Shanghai Shude Song, Yang Chen, Jin Zhu This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9101465/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Urban street accessibility for visually impaired individuals has traditionally been evaluated through the lens of barrier-free infrastructure. However, this study demonstrates that mobility stability is determined not merely by the physical presence of facilities, but by the extent to which environmental cues are coherently perceived, interpreted, and translated into action. Using Shanghai as a case study, we analyzed 309 questionnaire responses to examine how environmental barriers and psychological factors jointly shape travel intention. A random forest model achieved 78% classification accuracy in predicting travel avoidance behavior, identifying psychological anxiety and tactile paving obstruction as the most significant predictors. The findings suggest that disruptions in the built environment influence mobility primarily through the psychological mediation of environmental uncertainty. Furthermore, k-means clustering identified four distinct mobility profiles, designated as: Stable Traditional Users, Vulnerable Environment-Dependent Users, Technology-Assisted but Environmentally Sensitive Users, and Technology-Empowered Low-Anxiety Users. These groups exhibit significant variances in perceived environmental barriers, psychological anxiety, and assistive technology adoption. This study establishes a hierarchical framework linking environmental signals, perception, and behavior. The results reveal that mobility stability emerges through differentiated "interruption-compensation" pathways, advocating for a paradigm shift in accessibility governance from basic facility provision toward semantic reliability and digital-physical synchronization. Earth and environmental sciences/Environmental social sciences Biological sciences/Psychology Social science/Psychology Hierarchical environmental semantic reconstruction Visually impaired pedestrians Stable and coherent mobility Accessibility governance Urban streets Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Introduction Urban street environments in high-density cities present persistent challenges for visually impaired pedestrians. According to the World Health Organization 1 , more than 2.2 billion people worldwide experience varying degrees of visual impairment, a significant proportion of whom are affected by age-related visual pathologies 2 , 3 . As global urbanization and population aging accelerate, ensuring safe and independent mobility has become a cornerstone of inclusive urban development. Although accessibility standards have been progressively institutionalized, everyday sidewalk conditions frequently diverge from design intentions. In megacities like Shanghai, damaged tactile paving, sidewalk encroachment, and informal occupation disrupt navigational continuity. Under such conditions, the critical challenge shifts from the mere physical availability of infrastructure to its functional legibility—whether street environments provide coherent and interpretable cues that support autonomous travel in complex settings. For visually impaired individuals, mobility relies on non-visual channels, including tactile feedback, auditory signals, and spatial memory. Research indicates that spatial orientation depends on the coherence of environmental structures 4 , 5 . When environmental signals are fragmented, cognitive load increases, elevating psychological stress 6 , 7 . Furthermore, the effectiveness of assistive systems hinges not only on technological precision but on perceived transparency and user agency, which are central to trust formation 8 . Interventions that remain exclusively tech-centric or visually oriented may fail to reduce perceptual barriers and may even reinforce existing structural inequalities 9 , 10 . Tactile paving has long been regarded as foundational infrastructure for independent mobility. However, prior research indicates that streetscape characteristics and sidewalk conditions shape pedestrian movement patterns in nuanced ways 11 . For visually impaired pedestrians, the functionality of tactile paving depends less on its nominal installation than on its semantic continuity—whether it provides stable, predictable, and interpretable guidance across space. In many Chinese cities, dense pedestrian flows, non-motorized traffic, and insufficient maintenance contribute to damaged, blocked, or interrupted tactile paths 12 . These disruptions do not merely introduce physical hazards; they fragment environmental meaning. When tactile cues are unreliable, individuals may experience heightened anxiety, reduced confidence, and increased caution, potentially reshaping their willingness to travel independently. Parallel to improvements in physical infrastructure, smart mobility technologies—including GPS navigation, computer vision systems, and AI-based object recognition—have expanded rapidly 13 – 15 . These technologies typically operate at macro spatial scales, focusing on route calculation or obstacle detection. However, micro-scale street disruptions, temporary obstructions, and facility breakdowns often exceed their predictive capacity. Moreover, scholars have highlighted a persistent disconnect between technological innovation and the policy-driven needs of vulnerable populations 16 . Existing studies tend to evaluate either infrastructure provision or standalone assistive technologies, while fewer investigations integrate environmental conditions, psychological perception, and behavioral outcomes within a unified analytical framework. This fragmentation in existing research raises several important questions regarding the mobility of visually impaired pedestrians in complex street environments. In particular, it remains unclear how street-level environmental disruptions shape everyday mobility experiences, how psychological responses mediate the relationship between environmental conditions and travel intention, and how different user groups adapt to such disruptions through distinct mobility strategies. Addressing these questions requires moving beyond isolated evaluations of facilities or technologies toward a structural understanding of how perception, interpretation, and action interact in everyday urban contexts. To address these questions, this study examines travel intention among visually impaired pedestrians in Shanghai using questionnaire data and quantitative modeling. It identifies key environmental and psychological factors associated with travel avoidance and delineates differentiated mobility profiles across user groups. By integrating environmental perception, psychological mediation, and behavioral decision-making within a structured analytical framework, the study provides an empirically grounded understanding of mobility instability and offers insights for improving inclusive urban street environments. Literature Review & Theoretical Framework Structural Inequality and Environmental Discontinuity The mobility challenges faced by visually impaired individuals are closely linked to structural inequalities embedded in urban environments. Rather than arising solely from individual impairments, mobility barriers often result from mismatches between environmental design and non-visual perception. Research on smart cities increasingly emphasizes governance quality and social inclusion rather than technological efficiency alone. Batty et al. 17 argue that the value of smart urbanism lies in improving long-term public governance through systemic integration rather than short-term technological deployment. Similarly, Anarfi et al. 18 highlight that urban sustainability depends not only on infrastructure but also on citizens’ perceptions, values, and behavioral intentions. From the perspective of spatial cognition, Golledge 19 argues that spatial exclusion does not stem primarily from individual limitations but from the misalignment between environmental structures and perceptual modalities. Digital transformation alone cannot eliminate these mismatches and may even intensify them when technological systems assume visual interaction as the default mode. Van Dijk 20 further notes that the digital divide reflects broader structural inequalities rather than simple disparities in technological access. Urban governance research also points out that marginalized experiences are often insufficiently represented in digital infrastructures. Kitchin 21 , 22 shows that algorithmic governance may produce forms of “digital invisibility” when certain groups are excluded from data representation. In such contexts, fragmented environmental information can translate directly into navigational risk for visually impaired users. Taken together, these studies suggest that the key issue lies not only in physical accessibility but also in the discontinuity of environmental information that disrupts stable mobility experiences. Technological Assistance and the Detection-Action Disconnect Mobility assistance technologies for visually impaired users have evolved through several stages, each addressing particular challenges while also revealing new limitations. Traditional mobility tools, such as white canes and guide dogs, provide reliable tactile feedback for close-range navigation. However, they offer little anticipatory information and rely heavily on the continuity of tactile paving and surrounding environmental cues 19 . GPS-based navigation systems later introduced broader spatial guidance. Platforms such as the Drishti system 13 , enable integrated navigation services, yet their performance often declines in dense urban environments due to signal obstruction and positional drift. Coordinate-based guidance may also struggle to represent temporary obstacles or rapidly changing street conditions 23 . More recent approaches employ computer vision and artificial intelligence to recognize objects such as crosswalks and obstacles 14 . Although these systems significantly improve environmental detection, their reliability remains sensitive to lighting conditions, weather variability, and computational constraints. Frequent auditory prompts may also increase cognitive load during navigation 15 . Across these stages, technological development has primarily focused on detection and recognition. Far less attention has been paid to how environmental information is organized and translated into navigational action. This gap between information detection and behavioral response can be described as a detection-action disconnect, which limits the ability of existing systems to support stable mobility in complex street environments. Cognitive Mapping and the Hierarchy of Environmental Semantics Understanding urban environments requires more than sensory input; it also involves cognitive processes that organize spatial information into meaningful structures. Cognitive mapping theory provides an important framework for explaining how individuals construct spatial knowledge. Stea and Downs 24 describe cognitive maps as processes through which individuals acquire, encode, and organize environmental information. Similarly, Lynch 4 emphasizes that spatial environments must exhibit coherence and legibility in order to be easily understood. For visually impaired individuals, spatial cognition relies on the integration of tactile cues, auditory signals, and mental representations. Wayfinding therefore involves a sequence of processes including information acquisition, interpretation, and behavioral response. Studies show that blind individuals construct spatial knowledge through continuous perceptual experience 10 and that the stability of non-visual environmental cues directly influences spatial confidence and orientation 7 . Spatial meaning thus emerges through interpretation rather than through raw sensory input alone. However, many navigation technologies focus primarily on coordinates or object detection without clearly indicating how users should respond to the information provided. When environmental cues are fragmented or inconsistent, interpretive coherence can be disrupted, thereby increasing uncertainty and cognitive load during navigation. Behavioral Mediation and Travel Intention Mobility decisions are also shaped by psychological processes. The Theory of Planned Behavior explains behavioral intention through attitudes, subjective norms, and perceived behavioral control 25 . In uncertain environments, psychological responses such as anxiety and perceived insecurity may weaken perceived behavioral control and reduce the willingness to travel independently 6 . Empirical studies indicate that environmental uncertainty often influences mobility behavior indirectly through psychological appraisal rather than through physical barriers alone 7 . Integrating psychological variables into mobility analysis therefore helps explain how environmental conditions influence travel intention, particularly for visually impaired users whose mobility experiences are closely related to environmental reliability and perceived safety. Environmental Semantic Reconstruction and Research Hypotheses Although previous research has examined visually impaired mobility through perspectives such as spatial justice, cognitive mapping, and behavioral intention theory, an integrated explanation linking environmental perception, psychological mediation, and behavioral outcomes remains limited 26 – 28 . Many studies focus either on accessibility infrastructure or on assistive technologies while treating environmental conditions and behavioral responses as separate domains 29 , 30 . To address this gap, this study proposes a layered framework of environmental semantic reconstruction. Environmental semantics is defined as the structured organization of environmental meaning that enables individuals to translate perception into action (Fig. 1 ). This process operates across three interconnected layers: Perceptual Semantic Layer: the physical continuity and reliability of environmental cues Interpretive Semantic Layer: the cognitive integration of spatial information and the evaluation of environmental uncertainty Action-Oriented Semantic Layer: the translation of interpreted meaning into behavioral control and mobility decisions Environmental semantic reconstruction therefore refers to restoring coherence across these layers when discontinuities arise, ensuring alignment between perception, cognition, and action in complex urban environments. Based on this theoretical framework, this study proposes that environmental discontinuities generated by fragmented cues increase perceived uncertainty during navigation. Psychological responses, particularly perceived uncertainty and perceived behavioral control, are expected to mediate the relationship between environmental conditions and travel intention. Furthermore, different user groups may adopt distinct mobility strategies when responding to environmental disruptions, resulting in differentiated mobility outcomes. Methods Research Design This study adopts an empirical research design to investigate the mechanisms influencing mobility behavior among visually impaired pedestrians in urban street environments. The research integrates questionnaire-based data collection with quantitative modeling to examine the relationships between environmental conditions, psychological perception, and travel behavior. The analytical process consists of four sequential stages: empirical data collection, variable operationalization and preprocessing, quantitative modeling, and structural interpretation of mobility patterns. This stepwise design enables the identification of key influencing factors and heterogeneous mobility profiles within the visually impaired population. The overall research workflow, from empirical observation to framework development, is illustrated in Fig. 2 . Study Population The study focuses on visually impaired residents in Shanghai. Participants were recruited through collaboration with the Shanghai Association of the Blind and the Jing’an District Disabled Persons’ Federation. Data collection took place between December 1 and December 15, 2025, using the Tencent Survey online platform, which supports accessibility features such as screen reader compatibility and voice-assisted navigation. Respondents completed the questionnaire independently using their personal devices. A total of 309 valid responses were obtained. According to statistics from the Shanghai Disabled Persons’ Federation (2023), the registered visually impaired population in Shanghai is 95,946, meaning the sample represents approximately 0.32% of the population. User groups were not predefined during data collection. Instead, population heterogeneity was explored during the analytical stage using clustering techniques. Participation in this study was entirely voluntary. All respondents were fully informed of the study objectives and their right to withdraw at any time prior to completing the questionnaire. This research was conducted in accordance with the Declaration of Helsinki and was approved by the Science and Technology Ethics Committee of Donghua University (Approval Number: RLSSZYJ202603170025). The study was facilitated by the Shanghai and Jing’an District Disabled Persons’ Federations. Written informed consent was obtained from all participants. We further confirm that all participants were adults aged 18 and above, and all responses were anonymized to protect personal identifying information. Data Collection The data were derived from the Shanghai Public Space Experience Questionnaire for Visually Impaired Individuals, designed to capture environmental perception, facility experience, psychological responses, and travel behavior in urban street environments. To ensure accessibility for participants with different levels of vision, the questionnaire platform supported screen readers and voice-assisted navigation, enabling both blind and low-vision individuals to complete the survey independently. The questionnaire was structured around five dimensions: (A) Demographic Characteristics and Assistive Devices, (B) Perceived Environmental Barriers, (C/D) Facility Experience, (F) Psychological-Sensory Experience, and (E) Mobility Behavior Intention. Mobility intention represents the behavioral outcome in real street contexts, while the remaining dimensions capture environmental and psychological factors that may influence this outcome. A summary of questionnaire variables and their analytical roles is presented in Table 1 . Table 1 Overview of questionnaire dimensions and variable roles. Dimension Code Item description Scale Role in analysis Individual attributes A1 Level of visual impairment Categorical (Nominal) Control variable A2 Primary travel frequency Categorical (Multiple Choice) Independent Variable A3 Preferred assistive modes for smart systems Categorical (Multiple Choice) Design Reference Environmental barriers B1 Frequency of physical obstacles (obstruction/broken paving) Likert (1–5) Independent Variable B2 Impact of obstacles on safety and convenience Likert (1–5) Independent Variable B3 Subjective major difficulties (noise, landmarks, speed) Categorical (Multiple Choice) Independent Variable Navigation & Facility Experience C1 Convenience of facility usage in public spaces Likert (1–5) Mediating Variable C2 Difficulty in indoor spatial orientation and identification Likert (1–5) Mediating Variable D1 Dependence on physical and digital navigation aids Likert (1–5) Independent Variable D2 Overall satisfaction with current assistive methods Categorical (Multiple Choice) Outcome Variable / Feature Travel Intention & Behavior E1 Willingness to reduce or avoid outdoor travel Categorical (Ordinal) Dependent Variable E2 Primary reasons for travel avoidance Categorical (Multiple Choice) Dependent Variable Psychological Perception F1 Psychological anxiety in unfamiliar environments Likert (1–5) Mediating Variable F2 Perceived lack of safety during travel Likert (1–5) Mediating Variable F3 Specific psychological stressors/worries (e.g., collisions, getting lost) Categorical (Multiple Choice) Feature Engineering / Explainability Qualitative Support G1/G2 Overall satisfaction and open-ended suggestions Likert / Qualitative Qualitative Support During the distribution process, on-site briefing sessions were also conducted to explain the questionnaire and assist participants when necessary. Variable Operationalization and Data Preprocessing Questionnaire responses—primarily reflecting subjective perceptions, environmental experiences, and behavioral tendencies—were transformed into quantifiable variables suitable for statistical and machine learning analysis. During this process, each variable was operationalized to retain its conceptual meaning while ensuring numerical interpretability and analytical consistency. Based on the questionnaire structure, variables were organized into five conceptual dimensions: demographic characteristics and assistive device usage, perceived environmental barriers, facility and navigation experience, psychological perception, and travel behavior intention. These dimensions correspond to the analytical framework developed in this study and provide the basis for subsequent modeling. Prior to analysis, the raw survey data underwent systematic data cleaning and consistency verification. Responses were screened for logical consistency and valid response ranges, and incomplete or inconsistent entries were removed. Variables that were redundant, weakly related to the research objectives, or highly collinear variables were excluded during the feature screening stage to mitigate multicollinearity. After this process, a total of fifteen core features were retained as the primary explanatory variables for modeling. To prepare the dataset for machine learning analysis, several preprocessing procedures were applied. Continuous and ordinal variables were standardized using z-score normalization to ensure comparability across different measurement scales. Categorical variables were encoded as binary or dummy variables while preserving ordinal structure where applicable. All preprocessing and modeling procedures were implemented in Python (version 3.12). After preprocessing, the resulting dataset contained no missing values and was structured as a feature matrix suitable for machine learning modeling. For supervised learning analysis, the travel intention variable—specifically whether respondents would avoid certain street environments—was binarized as the target variable, while all remaining variables were used as input features. Analytical Strategy Using the structured dataset described above, this study developed a problem-oriented analytical framework to examine travel avoidance among visually impaired pedestrians. The objective extends beyond predictive accuracy and focuses on identifying structurally significant factors and population heterogeneity. Accordingly, the analysis combines supervised classification with unsupervised clustering. The analytical procedure is summarized in Fig. 3 . First, supervised classification models were used to examine relationships between environmental and psychological variables and travel avoidance behavior. Three representative models were compared: Logistic regression, which serves as a baseline model providing interpretable coefficients indicating the direction and magnitude of influence. Random forest (RF), an ensemble tree-based model capable of capturing nonlinear relationships and variable interactions while providing feature importance measures. Gradient boosting decision tree (GBDT), an iterative ensemble method that improves predictive performance by sequentially modeling residual errors. To ensure the generalizability of the results, all models were evaluated using k-fold cross-validation, mitigating the dependency on a single train-test split. Regularization techniques such as parameter tuning and tree-depth limitation were applied to reduce overfitting. Model performance was assessed using accuracy and F1-score, with the latter emphasized due to potential class imbalance. To ensure predictive reliability, random forest feature importance was derived using Gini impurity to quantify the contribution of environmental and psychological factors. For population heterogeneity, k-means clustering was optimized via the Elbow Method and Silhouette Coefficient to determine the most representative mobility profiles. Feature importance derived from the random forest model was further analyzed to identify variables with strong explanatory value. To explore heterogeneity within the sample, clustering analysis was conducted using key perceptual and behavioral variables, including environmental barrier perception, psychological burden, and assistive technology use. This analysis revealed distinct mobility profiles characterized by different combinations of perception, anxiety, and technology reliance. The integration of classification and clustering enables the identification of both key influencing factors and differentiated mobility patterns within the visually impaired population, providing the empirical foundation for the structural interpretation presented in the Results section. Results Based on the analytical framework and model specifications outlined above, this section presents the empirical results of the urban street mobility analysis for visually impaired individuals. The findings are organized into three components: model performance in predicting travel intention, identification of key influencing factors, and group differences within the sample. All analyses were conducted on the processed dataset following variable operationalization and preprocessing. Performance Evaluation of Travel Intention Prediction Models After data preprocessing and feature engineering, three models—logistic regression, random forest, and gradient boosting decision tree (GBDT)—were implemented to predict the binary outcome of whether individuals tend to reduce outdoor travel. Model performance was evaluated using five-fold cross-validation. As shown in Fig. 4 , the three models differ in predictive performance. Random forest achieves the highest overall accuracy (approximately 78%) and an F1-score close to 0.78, demonstrating balanced precision and recall in identifying individuals who reduce outdoor mobility. In contrast, logistic regression performs less effectively, suggesting limited ability to capture nonlinear relationships within the dataset. The GBDT model reaches accuracy levels comparable to random forest; however, its performance fluctuates more across validation folds, indicating slightly lower stability. Across all evaluation metrics, random forest consistently outperforms logistic regression and GBDT. Random forest was therefore selected as the primary model for subsequent feature importance analysis due to its superior predictive performance 31 . Feature Importance Analysis of Influencing Factors Following model selection, feature importance scores derived from the random forest model were examined to identify variables most strongly associated with reduced outdoor mobility (Fig. 5 ). The results indicate that psychological anxiety ranks highest among all predictors, highlighting its central role in travel avoidance behavior. This finding suggests that mobility decisions are shaped not only by objective environmental conditions but also by perceived uncertainty during movement. When environmental feedback lacks reliability and predictability, perceived safety diminishes, increasing the likelihood of avoiding independent travel. In addition to psychological anxiety, the frequency of occupied tactile paving and wayfinding difficulty also exhibit high importance scores. Occupied tactile paving reflects disruptions in environmental continuity, whereas wayfinding difficulty points to informational gaps when locating entrances, crossings, or key spatial nodes. Together, these variables suggest a pathway through which environmental disruptions elevate psychological stress, ultimately reducing willingness to travel. These results highlight the joint influence of environmental conditions and psychological perception on travel intention. User Profile Analysis Based on k-means Clustering Beyond identifying overall influencing factors, the study further explores heterogeneity within the visually impaired population through clustering analysis. Individuals were grouped using k-means clustering based on perceptual and behavioral characteristics. Clustering variables were selected according to both random forest feature importance results and theoretical considerations. Specifically, visual impairment level (A1), assistive device usage (A2), perceived environmental barriers (B1), and psychological anxiety (F1) were included as input features. Collectively, these variables capture environmental exposure, psychological state, and coping strategies during urban street travel. The optimal number of clusters was determined as K = 4 using the elbow method and silhouette coefficient evaluation. The clustering solution was visualized using principal component analysis (PCA), as shown in Fig. 6 . The four clusters display distinct configurations in psychological perception, environmental experience, and assistive strategy use. Stable Traditional Users (Cluster 1) Cluster 1 reports the lowest level of psychological anxiety ( M = 2.54, SD = 1.07) and minimal reliance on smartphone-based navigation ( M = 0.01, SD = 0.11). Perceived environmental barriers are also relatively low compared with other clusters. Members of this group primarily depend on traditional assistive strategies, such as white canes and accumulated spatial familiarity. The low anxiety level suggests well-developed coping mechanisms and adaptive strategies formed through long-term environmental experience. Rather than relying on digital systems, their mobility patterns reflect established spatial knowledge and confidence in conventional tactile and spatial cues. This pattern suggests that mobility in this group relies primarily on environmental familiarity rather than technological assistance. Vulnerable Environment-Dependent Users (Cluster 2) Cluster 2 exhibits the highest level of psychological anxiety ( M = 3.53, SD = 0.98) and the strongest perception of environmental barriers, including tactile paving obstruction and wayfinding difficulty, indicating that their elevated anxiety is closely coupled with frequent encounters with physical environmental disruptions. Despite facing substantial environmental challenges, smartphone usage remains relatively low ( M = 0.18, SD = 0.38), indicating limited technological compensation. The combination of high environmental exposure and low technological reliance characterizes a vulnerable mobility profile. Travel behavior in this group appears strongly constrained by environmental uncertainty, with relatively low reliance on technological assistance. Technology-Assisted but Environmentally Sensitive Users (Cluster 3) Cluster 3 shows elevated psychological anxiety ( M = 3.40, SD = 1.31) and high perception of environmental barriers, while demonstrating moderate adoption of smartphone-based navigation ( M = 0.53, SD = 0.50). Although assistive technology is actively used, environmental uncertainty continues to affect psychological stability and mobility confidence, suggestive of a persistent gap between digital information and physical reality. This indicates that while smartphone-based navigation provides supplementary information, it cannot fully mitigate the mobility risks induced by severe physical infrastructure discontinuities. Technology-Empowered Low-Anxiety Users (Cluster 4) Cluster 4 exhibits the highest smartphone usage rate ( M = 1.00, SD = 0.00) and comparatively low levels of anxiety ( M = 2.53, SD = 1.02). This mobility profile suggests strong technological adaptation and effective use of assistive navigation tools. In this group, technology appears to function as a stabilizing mechanism that reduces uncertainty and supports independent mobility. Statistical Validation of Cluster Differences To evaluate whether the four-cluster solution represents substantively distinct mobility profiles, a one-way analysis of variance (ANOVA) was conducted with cluster membership (K = 4) as the grouping variable. Five variables were selected for validation based on theoretical relevance and feature importance results: psychological anxiety, frequency of tactile paving obstruction, frequency of tactile paving defects, frequency of wayfinding difficulty, and smartphone usage rate for navigation. As shown in Table 2 , significant differences were observed across clusters for all variables ( p < 0.001). Effect sizes ( η² ) ranged from 0.159 to 0.646, indicating moderate to large between-group effects. Smartphone usage rate demonstrated the largest effect size ( η² = 0.646), underscoring the structurally differentiating role of technological adaptation across groups. Environmental barrier indicators and psychological anxiety also showed substantial effect sizes, indicating clear differences among the clusters. Table 2 One-way ANOVA results across the four mobility clusters. Variable Cluster 1 mean Cluster 2 mean Cluster 3 mean Cluster 4 mean F value p value η² Level of psychological anxiety 2.54 ± 1.07 3.53 ± 0.98 3.40 ± 1.31 2.53 ± 1.02 19.169 p < 0.001 0.159 Frequency of tactile paving obstruction 2.95 ± 0.90 4.27 ± 0.61 3.97 ± 0.86 3.44 ± 0.85 40.387 p < 0.001 0.284 Frequency of tactile paving defects 2.35 ± 0.74 3.72 ± 0.77 3.50 ± 0.98 2.84 ± 0.95 41.524 p < 0.001 0.29 Frequency of difficulty finding one's way 2.28 ± 0.79 3.63 ± 0.79 3.55 ± 0.75 2.82 ± 0.87 48.891 p < 0.001 0.325 Smartphone usage rate 0.01 ± 0.11 0.18 ± 0.38 0.53 ± 0.50 1.00 ± 0.00 185.238 p < 0.001 0.646 Tukey’s post hoc tests (Table 3 ) confirmed these distinct patterns: Clusters 2 and 3 reported significantly higher environmental barriers and psychological anxiety than Clusters 1 and 4 ( p < 0.001), while Cluster 4 showed the highest smartphone usage. Table 3 Tukey post hoc comparisons of significant differences across the four mobility clusters. Variable Comparison Mean Difference p value 95% CI Level of psychological anxiety C1 vs C2 0.99 < 0.001 [0.548, 1.431] C1 vs C3 0.854 < 0.001 [0.374, 1.335] C2 vs C4 −1.004 < 0.001 [− 1.438, − 0.570] C3 vs C4 −0.869 < 0.001 [− 1.342, − 0.395] Frequency of tactile paving obstruction C1 vs C2 1.314 < 0.001 [0.984, 1.644] C1 vs C4 0.486 < 0.01 [0.166, 0.806] C2 vs C4 −0.828 < 0.001 [− 1.152, − 0.503] C3 vs C4 −0.527 < 0.01 [− 0.881, − 0.173] Frequency of tactile paving defects C1 vs C2 1.372 < 0.001 [1.023, 1.721] C1 vs C4 0.493 < 0.01 [0.155, 0.832] C2 vs C4 −0.879 < 0.001 [− 1.222, − 0.536] C3 vs C4 −0.657 < 0.001 [− 1.032, − 0.283] Frequency of wayfinding difficulty C1 vs C2 1.356 < 0.001 [1.028, 1.683] C1 vs C4 0.543 < 0.001 [0.225, 0.861] C2 vs C4 −0.813 < 0.001 [− 1.135, − 0.491] C3 vs C4 −0.732 < 0.001 [− 1.083, − 0.380] Smartphone usage rate C1 vs C2 0.165 < 0.01 [0.044, 0.286] C1 vs C3 0.522 < 0.001 [0.391, 0.654] C1 vs C4 0.988 < 0.001 [0.871, 1.105] C2 vs C3 0.357 < 0.001 [0.225, 0.490] C2 vs C4 0.823 < 0.001 [0.704, 0.942] C3 vs C4 0.466 < 0.001 [0.336, 0.595] These differences are visually summarized in Fig. 7 , which presents the mean values of the five validation variables across clusters. Clusters 2 and 3 form a high-barrier, high-anxiety profile, while Cluster 4 is characterized by strong technological adaptation despite moderate environmental exposure. Cluster 1 represents a relatively low-barrier, low-anxiety profile with limited reliance on digital assistance. Taken together, the ANOVA and post hoc analyses confirm that the clustering solution reflects statistically distinguishable mobility profiles rather than arbitrary algorithmic segmentation. The integration of clustering and statistical validation strengthens confidence that the identified profiles represent structurally differentiated mobility patterns. The four groups differ systematically in perceived environmental barriers, psychological response, and assistive behavior, providing a robust empirical basis for subsequent strategy differentiation. While clustering identifies statistically distinct behavioral configurations, the observed differences also reflect variations in environmental perception and psychological response among user groups. These empirical patterns provide the basis for the environmental semantic reconstruction framework introduced in the following section. Discussion From Behavioral Clusters to Differences in Environmental Semantics This study demonstrates that mobility stability depends on the continuity of environmental meaning—perceived intelligibility and predictability—rather than mere infrastructure presence. Unlike conventional research treating accessibility as physical provision, our findings reveal that mobility relies on whether information can be reliably perceived, cognitively integrated, and translated into action. This aligns with research emphasizing perceived accessibility in shaping behavior 32 , 33 . The four statistically distinct clusters identified in the Results reflect not merely behavioral variation, but differentiated pathways through which environmental meaning is constructed and stabilized. Importantly, these clusters do not represent differences in capability. They reflect differences in semantic interpretation pathways. Urban streets function not as neutral information carriers, but as environments whose meaning varies depending on cognitive habits and technological engagement. Although the analysis primarily relies on perception-based questionnaire data, many of the measured variables correspond to observable environmental conditions, such as tactile paving obstruction, facility defects, and navigation reliability. The findings therefore reflect the interaction between objective environmental cues and subjective interpretation, rather than subjective perception alone. Linking Behavioral Clusters to Environmental Reconstruction Strategies Behavioral clusters reveal distinct semantic disruptions manifesting from infrastructure failures. According to Lynch 4 , legibility is essential for navigation; for visually impaired pedestrians, this relies on consistent tactile paving and auditory signals. To bridge behaviors with infrastructure, we categorize elements by functional role: tactile paving and curb ramps form the perceptual-layer (primary signals), while auditory signals and beacons constitute the interpretive-layer (supplementary information). The four mobility clusters identified in this study can therefore be interpreted as reflecting different patterns of environmental cue disruption (as quantified in Table 2 ): Cluster 1 (Stable Traditional Users) exhibits relatively low perceived environmental barriers and anxiety, indicating that these individuals can still rely on the existing “legibility” of the environment. Their mobility suggests that current street infrastructure supports navigation when tactile paths remain continuous and predictable. For this group, environmental reconstruction involves preventive maintenance to ensure that the existing semantic structure does not degrade. Cluster 2 (Vulnerable Environment-Dependent Users) represents a perceptual-layer disruption. The random forest analysis (Fig. 5 ) and the significantly higher mean scores for tactile paving issues in Table 2 provide empirical evidence for this: the high importance scores for "tactile paving obstruction" and “defects” demonstrate that when these specific physical objects fail, the entire perceptual system of the pedestrian collapses, confirming that physical defects are the primary external triggers for the top-ranked psychological anxiety (Mean = 3.53, Table 2 ) observed in Fig. 5 . This statistical coupling between physical object failure, psychological response, and behavioral outcome validates the identification of these facilities as critical perceptual-layer nodes. Their high anxiety reflects a physical breakdown in the environment’s primary communication system. When the tactile “language” of the street is interrupted by obstacles, the environment becomes semantically “silent” or “misleading.” Consequently, stabilizing tactile information through systematic maintenance is the most direct intervention to restore environmental predictability. Cluster 3 (Technology-Assisted but Environmentally Sensitive Users) highlights an interpretive-layer mismatch. This pattern suggests that while digital navigation tools provide supplementary information, they often fail to align with physical environmental cues. According to James J. Gibson’s 34 ecological perception framework, navigation emerges from environmental affordances. If digital guidance does not correspond to physical tactile paving or crossing signals, the “information mismatch” undermines confidence. Reconstruction for this group must focus on the semantic integration of digital and physical layers. Specifically, this involves deploying Bluetooth beacons or RFID markers that trigger real-time auditory descriptions of intersection geometries to ensure they match the existing physical tactile cues. Cluster 4 (Technology-Empowered Low-Anxiety Users) demonstrates the potential of multi-modal environmental information. Their lower anxiety levels suggest that a redundant information system (physical + digital) can compensate for certain environmental limitations. For these users, reconstruction involves enhancing the coordination between physical infrastructure and digital data to create a seamless, multi-layered environmental semantic network.Such a network requires the high-precision coordination between physical sidewalk integrity and dynamic digital data, ensuring that real-time hazard alerts are seamlessly integrated with the stable infrastructure base. In summary, these behavioral clusters serve as diagnostic indicators of environmental performance. Rather than simply representing user preferences, they reveal where and how the urban street fabric fails to provide the necessary semantic cues for stable mobility. Theoretical and Mechanism Verification Although lacking VR testing, the identified links between cue disruption and anxiety align with Lynch’s 4 legibility and Gibson’s 34 affordance theories. This convergence of machine learning evidence and spatial theory validates the hierarchical framework, transforming pedestrian feedback into objective performance indicators. The convergence of machine learning evidence and spatial perception theory provides a robust basis for the proposed hierarchical reconstruction. This approach aligns with recent calls for hierarchical strategies in tactile paving deployment 35 and mirrors the broader transition in inclusive governance from physical provision toward technologically-integrated information reliability 36 . By transforming subjective pedestrian feedback into objective infrastructure performance indicators, this framework acts as a structural interpretation of established mechanisms. A Multihierarchical Environmental Semantic Reconstruction Framework Building on these differentiated patterns, environmental semantic reconstruction can be conceptualized as a multihierarchical framework. Mobility stability emerges from continuity across three interdependent layers: perceptual acquisition, interpretive integration, and action translation. The Perceptual Layer concerns the acquisition of environmental cues, including tactile continuity and spatial structure. Instability at this level initiates semantic breakdown. The Interpretive Layer involves cognitive integration and uncertainty evaluation. Disruption manifests as anxiety and reduced confidence, even where partial cues are present. The Action-Oriented Layer translates interpreted information into executable movement. Technological systems often intervene here through route confirmation and real-time adjustment feedback. Mobility stability is therefore not a direct function of infrastructure provision or technological presence alone. It depends on whether semantic continuity is maintained—or reconstructed—across these layers. The decisive variable across clusters is therefore not the intensity of technological assistance, but the layer at which reconstruction intervenes. Early-layer stabilization generates organic continuity; later-layer compensation produces mediated stability. Where no effective intervention exists, instability propagates. Environmental semantic reconstruction is thus defined as the structural reorganization of environmental meaning to restore coherence between perception, cognition, and action. Operationally, reconstruction involves: 1. Restoring perceptual continuity in street infrastructure; 2. Enhancing interpretive predictability in spatial cognition; 3. Reinforcing action-level confirmation mechanisms. This layered model clarifies how differentiated mobility profiles correspond to distinct interruption points and intervention priorities (Fig. 8 , Fig. 9 ). From infrastructure provision to semantic continuity in accessibility governance Governance must shift from facility expansion to ensuring semantic reliability and synchronization. At the perceptual level, metrics should prioritize obstruction frequency and maintenance response over mere installation rates, institutionalizing enforcement against sidewalk occupation. At the interpretive level, inter-departmental coordination is essential to synchronize infrastructure upgrades and reduce environmental fragmentation. At the action level, digital systems should serve as confirmation mechanisms; thus, data accuracy and synchronization with physical conditions are critical. Differentiated cluster needs demand targeted priorities: environmentally vulnerable users require perceptual stabilization, while technology-dependent groups need digital-physical integration and data reliability. Moving beyond the “average user” assumption, reframing accessibility as semantic stability promotes a continuity-oriented model of inclusive street management. By empirically linking perceptual uncertainty with behavioral adaptation, this study provides a framework for data-informed, inclusive urban governance. Conclusion Although the empirical data were collected in Shanghai, the environmental semantic reconstruction framework proposed in this study is conceptually transferable to other high-density urban contexts facing comparable accessibility challenges. By integrating random forest modeling, k-means clustering, and statistical validation, this study provides a data-driven examination of mobility intentions among visually impaired pedestrians. The findings indicate that mobility stability in urban street environments is not determined solely by the mere provision of accessibility infrastructure. Rather, it emerges from the continuity—or breakdown—of environmental meaning across perceptual, interpretive, and action-oriented layers of navigation. Even in cities where tactile paving systems and barrier-free facilities are formally implemented, discontinuous environmental cues and unreliable infrastructure performance can generate psychological anxiety and erode mobility confidence. These findings suggest that accessibility research and governance have historically prioritized physical installation while insufficiently addressing the semiotic and cognitive dimensions of the lived environmental experience. By identifying four differentiated user clusters, this study demonstrates that mobility stability develops through distinct environmental disruption-behavioral compensation pathways. For some individuals, accumulated environmental familiarity sustains perceptual continuity. For others, technological mediation at the action layer partially or fully compensates for perceptual disruption. Mobility outcomes are therefore shaped not by the intensity of technological deployment per se, but by the level within the perceptual-interpretive-action chain at which semantic reconstruction intervenes. From a governance perspective, the results highlight the need to complement infrastructure provision with reliability-oriented management of accessibility systems. Rather than focusing solely on the installation of barrier-free facilities, governance frameworks should consider the everyday usability and environmental consistency of accessibility infrastructure as central performance indicators. Within this perspective, the environmental semantic reconstruction framework provides a conceptual basis for integrating physical design, digital sensing, and adaptive governance mechanisms. Reconstruction therefore involves not merely increasing the amount of environmental information, but ensuring that environmental cues remain interpretable and dependable throughout the perceptual-interpretive-action chain. Emerging smart-city technologies, such as embedded sensing systems for monitoring tactile paving conditions and obstruction patterns, may support this objective by enabling data-informed maintenance and responsive management of urban accessibility infrastructure 37 , 38 . Importantly, such technological interventions should be aligned with differentiated user profiles so that reconstruction efforts address the specific layer at which semantic disruption occurs. Several limitations warrant consideration. The study is constrained by its sampling scope and single-city context. Future research could incorporate longitudinal field validation, cross-city comparative analysis, and prototype implementation testing to further examine how layered semantic reconstruction influences long-term behavioral stability and psychological adaptation. A pilot smart tactile paving system currently under development in Shanghai seeks to operationalize the framework proposed here. Systematic evaluation of this deployment will provide empirical evidence for refining the interaction among environmental reliability, technological mediation, and user experience. More broadly, the framework may be extended to other vulnerable populations, including older adults and individuals with mobility impairments, to assess its applicability within wider inclusive urban governance contexts. Declarations Funding This research was supported by the Shanghai Philosophy and Social Sciences Planning Project (Grant No. 2024BCK012), funded by the Shanghai Philosophy and Social Sciences Planning Office (Project title: “Research on the Spatial Agglomeration and Symbiotic Development Model of Shanghai’s Digital Creative Industry”). Author contributions S.S. and Y.C. contributed equally to this work. S.S. conceived and designed the research and supervised the study. Y.C. performed the data analysis and drafted the manuscript. J.Z. supervised the research and revised the manuscript. All authors reviewed and approved the final manuscript. Competing interests The authors declare no competing interests. Data availability The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request. Ethics approval and consent to participate This study was conducted in accordance with the Declaration of Helsinki. The research protocol and questionnaire were reviewed and approved by the Science and Technology Ethics Committee of Donghua University (Approval Number: RLSSZYJ202603170025), with additional administrative support and review by the Shanghai Disabled Persons’ Federation and the Jing’an District Disabled Persons’ Federation. Written informed consent was obtained from all participants prior to their involvement. We confirm that all participants were adults aged 18 and above; no minors were involved in this research. All data were fully anonymized before analysis to ensure privacy. References World Health Organization. World Report on Vision (Geneva, 2019). GBD 2020 Blindness and Vision Impairment Collaborators. Causes of blindness and vision impairment in 2020 and trends over 30 years, and prevalence of avoidable blindness in relation to VISION 2020: the Right to Sight: an analysis for the Global Burden of Disease Study. Lancet Global Health . 9 , e144–e160 (2021). Zou, M. et al. Prevalence of visual impairment among older Chinese population: A systematic review and meta-analysis. J. Glob Health . 11 , 08004 (2021). Kevin Lynch. The Image of the City (The MIT, 1960). Passini, R. Wayfinding design: logic, application and some thoughts on universality. Des. Stud. 17 , 319–331 (1996). Hersh, M. & Deafblind People Communication, Independence, and Isolation. J Deaf Stud. Deaf Educ. 18 , 446–463 (2013). Gori, M., Cappagli, G., Baud-Bovy, G. & Finocchietti, S. Shape Perception and Navigation in Blind Adults. Front Psychol 8 , (2017). Liu, Y., Li, H. & Pan, Y. Understanding perceived ride safety and trust formation in robotaxi services under day and night conditions. SCIENTIFIC REPORTS 15 (2025). UN-Habitat. The Value of Sustainable Urbanization. (2020). Schinazi, V. R., Thrash, T. & Chebat, D. Spatial navigation by congenitally blind individuals. Wires Cogn. Sci. 7 , 37–58 (2016). Zhang, Z., Sun, T., Fisher, T. & Wang, H. The relationships between the campus built environment and walking activity. SCIENTIFIC REPORTS 14 (2024). Zhou, K., Hu, C., Zhang, H., Hu, Y. & Xie, B. Why do we hardly see people with visual impairments in the street? A case study of Changsha, China. Appl. Geogr. 110 , 102043 (2019). Ran, L., Helal, S. & Moore, S. Drishti: an integrated indoor/outdoor blind navigation system and service. in Second IEEE Annual Conference on Pervasive Computing and Communications , Proceedings of the 23–30 (IEEE, Orlando, FL, USA, 2004). (2004). 10.1109/PERCOM.2004.1276842 Mascetti, S., Ahmetovic, D., Gerino, A., Bernareggi, C. & ZebraRecognizer Pedestrian crossing recognition for people with visual impairment or blindness. Pattern Recogn. 60 , 405–419 (2016). Bhowmick, A. & Hazarika, S. M. An insight into assistive technology for the visually impaired and blind people: state-of-the-art and future trends. J. Multimodal User Interfaces . 11 , 149–172 (2017). Daovisan, H. Responsible or Sustainable AI? Circular Economy Models in Smart Cities. Sustainability 18 , 398 (2025). Batty, M. Inventing Future Cities . (2018). https://doi.org/10.7551/mitpress/11923.001.0001 Kwasi Anarfi, C., Shiel & Ross, A. Hill. The public’s perspectives of urban sustainability: A comparative analysis of two urban areas in Ghana. Habitat Int. 111 , 102496 (2021). Reginald, G. & Golledge Geography and the disabled: A survey with special reference to vision impaired and blind populations. Trans. Inst. Br. Geogr. 18 , 63–85 (1993). Dijk, J. The Digital Divide (Polity, 2020). 10.1002/asi.24355 Kitchin, R. The Real Time City༟Big data and smart urbanism. GeoJournal 79, 1–14 (2014). Rob Kitchin. Data-driven Urbanism. Data City . 13 https://doi.org/10.4324/9781315407388 (2017). Velázquez, R. et al. An Outdoor Navigation System for Blind Pedestrians Using GPS and Tactile-Foot Feedback. Appl. Sci. 8 , 578 (2018). David Stea. Image and Environment Cognitive Mapping and Spatial Behavior (Routledge, 2017). https://doi.org/10.4324/9780203789155 Ajzen, I. The theory of planned behavior. Organ. Behav. Hum Decis. Process. 50 , 179–211 (1991). El-taher, F. E., Taha, A., Courtney, J. & Mckeever, S. A Systematic Review of Urban Navigation Systems for Visually Impaired People. Sensors 21 , 3103 (2021). Imrie, R. & Luck, R. Designing inclusive environments: rehabilitating the body and the relevance of universal design. Disabil. Rehabil . 36 , 1315–1319 (2014). Lättman, K., Olsson, L., Friman, M. & Fujii, S. Perceived Accessibility, Satisfaction with Daily Travel, and Life Satisfaction among the Elderly. IJERPH 16 , 4498 (2019). Imrie, R. & Hall, P. InclusiveDesign Designinganddevelopingaccessibleenvironments (Spon, 2001). 10.4324/9780203362501 Marion, A., Hersh & Johnson, M. A. Assistive Technology for Visually Impaired and Blind People Springer London,. (2008). https://doi.org/10.1007/978-1-84628-867-8 Liaw, A. & Wiener, M. Classification and Regression by randomForest. R News . 2 , 18–22 (2002). Curl, A., Nelson, J. D. & Anable, J. Does Accessibility Planning address what matters? A review of current practice and practitioner perspectives. Res. Transp. Bus. Manage. 2 , 3–11 (2011). Negm, H., De Vos, J. & Pot, F. El-Geneidy, A. Perceived accessibility: A literature review. J. Transp. Geogr. 125 , 104212 (2025). James, J. & Gibson The Ecological Approach to Visual Perception New York,. (2014). https://doi.org/10.4324/9781315740218 Fadhlillah, F. Rethinking the Tactile Paving Installation System Based on the City Rhythm of Visually Impaired Pedestrians in Urban Networks. Jurnal Pembangunan Wilayah dan. Kota 20 , (2024). Almoshaogeh, M. et al. A Review on Disability-Inclusive Public Transportation: Current Barriers and Prospects. IEEE Access. 13 , 75769–75786 (2025). Caragliu, A. & Del Bo, C. F. Smart innovative cities: The impact of Smart City policies on urban innovation. Technol. Forecast. Soc. Chang. 142 , 373–383 (2019). United Nations, General Assembly. New Urban Agenda . (2016). https://unhabitat.org/sites/default/files/2019/05/nua-english.pdf Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted 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-9101465","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":616661777,"identity":"e0d84439-e31a-4776-80e8-a54755072f0a","order_by":0,"name":"Shude Song","email":"","orcid":"","institution":"Donghua University","correspondingAuthor":false,"prefix":"","firstName":"Shude","middleName":"","lastName":"Song","suffix":""},{"id":616661778,"identity":"4171d085-0726-4163-9f2a-2b2b5619c7cb","order_by":1,"name":"Yang Chen","email":"","orcid":"","institution":"Donghua University","correspondingAuthor":false,"prefix":"","firstName":"Yang","middleName":"","lastName":"Chen","suffix":""},{"id":616661779,"identity":"526b683f-b01d-4be2-8bd3-53b838fc6ea4","order_by":2,"name":"Jin Zhu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAwUlEQVRIiWNgGAWjYDACCQYGgwQGG8YNIA4PCVrSSNQCBIdJ0CI/u8eg4OGO87LbJRIYH7xtY5A3J6SFcc4ZA4PEM7eNd85IYDac28ZguLOBgBZmiRyglrbbiRtuJLBJ87YxJBgcIKCFDaLlHEgL+2+itPBAtBwA28JMlBYJibQCoJZk4w1nHjZLzjknYbiBkBb5GcnbDH+22cluOJ588MObMht5graAvGMAoRkbGGDRRAgwPyBK2SgYBaNgFIxcAAAN0j/iLISwigAAAABJRU5ErkJggg==","orcid":"","institution":"Donghua University","correspondingAuthor":true,"prefix":"","firstName":"Jin","middleName":"","lastName":"Zhu","suffix":""}],"badges":[],"createdAt":"2026-03-12 07:25:36","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9101465/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9101465/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":106189721,"identity":"7d55db6b-3636-4989-bb12-291222f1177e","added_by":"auto","created_at":"2026-04-05 17:10:57","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":174415,"visible":true,"origin":"","legend":"\u003cp\u003eMulti-Layer Environmental Semantic Structure and Analytical Logic.\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9101465/v1/51ebe480d7c9a247c1df4804.jpeg"},{"id":106402934,"identity":"1231e0d1-8d12-4351-b144-81440fbc81f7","added_by":"auto","created_at":"2026-04-08 09:13:13","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":424283,"visible":true,"origin":"","legend":"\u003cp\u003eResearch workflow from empirical data collection to environmental semantic reconstruction framework.\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9101465/v1/7509c68e897419c373d6acd7.jpeg"},{"id":106402677,"identity":"0f169c2b-afa2-48a2-8e9c-cca9ea28d174","added_by":"auto","created_at":"2026-04-08 09:12:32","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":124760,"visible":true,"origin":"","legend":"\u003cp\u003eAnalytical Pathway from Empirical Data to Structural Interpretation.\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9101465/v1/51375ef30b0f4c0dfb183f37.jpeg"},{"id":106189723,"identity":"fb1141d5-1b22-4a94-8651-ccc534625c48","added_by":"auto","created_at":"2026-04-05 17:10:57","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":86819,"visible":true,"origin":"","legend":"\u003cp\u003ePerformance comparison of machine learning models for predicting reduced outdoor mobility.\u003c/p\u003e","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9101465/v1/f4639736b31b152c6687a71c.jpeg"},{"id":106403018,"identity":"603f2a5a-cc02-47b7-85d7-c306fa757f1f","added_by":"auto","created_at":"2026-04-08 09:13:24","extension":"jpeg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":173421,"visible":true,"origin":"","legend":"\u003cp\u003eFeature importance of environmental and psychological factors influencing reduced outdoor mobility among visually impaired pedestrians.\u003c/p\u003e","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9101465/v1/973a768976cf82801915ab2e.jpeg"},{"id":106189724,"identity":"6c91a5fa-6915-428d-b475-1528be3844bc","added_by":"auto","created_at":"2026-04-05 17:10:57","extension":"jpeg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":195409,"visible":true,"origin":"","legend":"\u003cp\u003eK-means clustering visualization of visually impaired pedestrians (K = 4).\u003c/p\u003e","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9101465/v1/8aaafceb432a97306a05ccce.jpeg"},{"id":106403039,"identity":"4ecff6d3-cbdc-4bb9-85aa-d745c66a77c3","added_by":"auto","created_at":"2026-04-08 09:13:26","extension":"jpeg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":283643,"visible":true,"origin":"","legend":"\u003cp\u003eMean differences across clusters in psychological, environmental, and technological dimensions.\u003c/p\u003e","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9101465/v1/97824e0a31cf22d6cf99407c.jpeg"},{"id":106189727,"identity":"11b629db-1fdc-4eda-b96a-f6c141920515","added_by":"auto","created_at":"2026-04-05 17:10:57","extension":"jpeg","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":224319,"visible":true,"origin":"","legend":"\u003cp\u003eEnvironmental Semantic Reconstruction Pathways for Differentiated Mobility Profiles.\u003c/p\u003e","description":"","filename":"floatimage8.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9101465/v1/9e4b2bf30e3a343aab4a75df.jpeg"},{"id":106189726,"identity":"d29e1f5b-7344-4cb9-8c72-11ae70fcecf7","added_by":"auto","created_at":"2026-04-05 17:10:57","extension":"jpeg","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":476422,"visible":true,"origin":"","legend":"\u003cp\u003eIntegrated Infrastructure System for Environmental Semantic Continuity.\u003c/p\u003e","description":"","filename":"floatimage9.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9101465/v1/45e04a0d67f9e7fc8fb77b3d.jpeg"},{"id":109067425,"identity":"3fcd7342-ea10-4815-81f4-f112cb23fa20","added_by":"auto","created_at":"2026-05-12 09:48:10","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2655041,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9101465/v1/0fcd356f-f136-46fe-a4ef-6ddb3fc6b4a0.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Hierarchical Environmental Semantic Reconstruction for Stable Mobility of Visually Impaired Pedestrians on Urban Streets in Shanghai","fulltext":[{"header":"Introduction","content":"\u003cp\u003eUrban street environments in high-density cities present persistent challenges for visually impaired pedestrians. According to the World Health Organization\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e, more than 2.2\u0026nbsp;billion people worldwide experience varying degrees of visual impairment, a significant proportion of whom are affected by age-related visual pathologies\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e,\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. As global urbanization and population aging accelerate, ensuring safe and independent mobility has become a cornerstone of inclusive urban development.\u003c/p\u003e\u003cp\u003eAlthough accessibility standards have been progressively institutionalized, everyday sidewalk conditions frequently diverge from design intentions. In megacities like Shanghai, damaged tactile paving, sidewalk encroachment, and informal occupation disrupt navigational continuity. Under such conditions, the critical challenge shifts from the mere physical availability of infrastructure to its functional legibility\u0026mdash;whether street environments provide coherent and interpretable cues that support autonomous travel in complex settings.\u003c/p\u003e\u003cp\u003eFor visually impaired individuals, mobility relies on non-visual channels, including tactile feedback, auditory signals, and spatial memory. Research indicates that spatial orientation depends on the coherence of environmental structures\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e,\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e. When environmental signals are fragmented, cognitive load increases, elevating psychological stress\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e,\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e. Furthermore, the effectiveness of assistive systems hinges not only on technological precision but on perceived transparency and user agency, which are central to trust formation\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. Interventions that remain exclusively tech-centric or visually oriented may fail to reduce perceptual barriers and may even reinforce existing structural inequalities\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eTactile paving has long been regarded as foundational infrastructure for independent mobility. However, prior research indicates that streetscape characteristics and sidewalk conditions shape pedestrian movement patterns in nuanced ways\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e. For visually impaired pedestrians, the functionality of tactile paving depends less on its nominal installation than on its semantic continuity\u0026mdash;whether it provides stable, predictable, and interpretable guidance across space. In many Chinese cities, dense pedestrian flows, non-motorized traffic, and insufficient maintenance contribute to damaged, blocked, or interrupted tactile paths\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e. These disruptions do not merely introduce physical hazards; they fragment environmental meaning. When tactile cues are unreliable, individuals may experience heightened anxiety, reduced confidence, and increased caution, potentially reshaping their willingness to travel independently.\u003c/p\u003e\u003cp\u003eParallel to improvements in physical infrastructure, smart mobility technologies\u0026mdash;including GPS navigation, computer vision systems, and AI-based object recognition\u0026mdash;have expanded rapidly\u003csup\u003e\u003cspan additionalcitationids=\"CR14\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e. These technologies typically operate at macro spatial scales, focusing on route calculation or obstacle detection. However, micro-scale street disruptions, temporary obstructions, and facility breakdowns often exceed their predictive capacity. Moreover, scholars have highlighted a persistent disconnect between technological innovation and the policy-driven needs of vulnerable populations\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. Existing studies tend to evaluate either infrastructure provision or standalone assistive technologies, while fewer investigations integrate environmental conditions, psychological perception, and behavioral outcomes within a unified analytical framework.\u003c/p\u003e\u003cp\u003eThis fragmentation in existing research raises several important questions regarding the mobility of visually impaired pedestrians in complex street environments. In particular, it remains unclear how street-level environmental disruptions shape everyday mobility experiences, how psychological responses mediate the relationship between environmental conditions and travel intention, and how different user groups adapt to such disruptions through distinct mobility strategies. Addressing these questions requires moving beyond isolated evaluations of facilities or technologies toward a structural understanding of how perception, interpretation, and action interact in everyday urban contexts.\u003c/p\u003e\u003cp\u003eTo address these questions, this study examines travel intention among visually impaired pedestrians in Shanghai using questionnaire data and quantitative modeling. It identifies key environmental and psychological factors associated with travel avoidance and delineates differentiated mobility profiles across user groups. By integrating environmental perception, psychological mediation, and behavioral decision-making within a structured analytical framework, the study provides an empirically grounded understanding of mobility instability and offers insights for improving inclusive urban street environments.\u003c/p\u003e"},{"header":"Literature Review \u0026 Theoretical Framework","content":"\u003cp\u003eStructural Inequality and Environmental Discontinuity\u003c/p\u003e \u003cp\u003eThe mobility challenges faced by visually impaired individuals are closely linked to structural inequalities embedded in urban environments. Rather than arising solely from individual impairments, mobility barriers often result from mismatches between environmental design and non-visual perception.\u003c/p\u003e \u003cp\u003eResearch on smart cities increasingly emphasizes governance quality and social inclusion rather than technological efficiency alone. Batty et al.\u003csup\u003e17\u003c/sup\u003e argue that the value of smart urbanism lies in improving long-term public governance through systemic integration rather than short-term technological deployment. Similarly, Anarfi et al.\u003csup\u003e18\u003c/sup\u003e highlight that urban sustainability depends not only on infrastructure but also on citizens\u0026rsquo; perceptions, values, and behavioral intentions.\u003c/p\u003e \u003cp\u003eFrom the perspective of spatial cognition, Golledge\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e argues that spatial exclusion does not stem primarily from individual limitations but from the misalignment between environmental structures and perceptual modalities. Digital transformation alone cannot eliminate these mismatches and may even intensify them when technological systems assume visual interaction as the default mode. Van Dijk\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e further notes that the digital divide reflects broader structural inequalities rather than simple disparities in technological access.\u003c/p\u003e \u003cp\u003eUrban governance research also points out that marginalized experiences are often insufficiently represented in digital infrastructures. Kitchin\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e,\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e shows that algorithmic governance may produce forms of \u0026ldquo;digital invisibility\u0026rdquo; when certain groups are excluded from data representation. In such contexts, fragmented environmental information can translate directly into navigational risk for visually impaired users.\u003c/p\u003e \u003cp\u003eTaken together, these studies suggest that the key issue lies not only in physical accessibility but also in the discontinuity of environmental information that disrupts stable mobility experiences.\u003c/p\u003e \u003cp\u003eTechnological Assistance and the Detection-Action Disconnect\u003c/p\u003e \u003cp\u003eMobility assistance technologies for visually impaired users have evolved through several stages, each addressing particular challenges while also revealing new limitations.\u003c/p\u003e \u003cp\u003eTraditional mobility tools, such as white canes and guide dogs, provide reliable tactile feedback for close-range navigation. However, they offer little anticipatory information and rely heavily on the continuity of tactile paving and surrounding environmental cues\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eGPS-based navigation systems later introduced broader spatial guidance. Platforms such as the Drishti system\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e, enable integrated navigation services, yet their performance often declines in dense urban environments due to signal obstruction and positional drift. Coordinate-based guidance may also struggle to represent temporary obstacles or rapidly changing street conditions\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eMore recent approaches employ computer vision and artificial intelligence to recognize objects such as crosswalks and obstacles\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Although these systems significantly improve environmental detection, their reliability remains sensitive to lighting conditions, weather variability, and computational constraints. Frequent auditory prompts may also increase cognitive load during navigation\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eAcross these stages, technological development has primarily focused on detection and recognition. Far less attention has been paid to how environmental information is organized and translated into navigational action. This gap between information detection and behavioral response can be described as a detection-action disconnect, which limits the ability of existing systems to support stable mobility in complex street environments.\u003c/p\u003e \u003cp\u003eCognitive Mapping and the Hierarchy of Environmental Semantics\u003c/p\u003e \u003cp\u003eUnderstanding urban environments requires more than sensory input; it also involves cognitive processes that organize spatial information into meaningful structures. Cognitive mapping theory provides an important framework for explaining how individuals construct spatial knowledge.\u003c/p\u003e \u003cp\u003eStea and Downs\u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e describe cognitive maps as processes through which individuals acquire, encode, and organize environmental information. Similarly, Lynch\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e emphasizes that spatial environments must exhibit coherence and legibility in order to be easily understood.\u003c/p\u003e \u003cp\u003eFor visually impaired individuals, spatial cognition relies on the integration of tactile cues, auditory signals, and mental representations. Wayfinding therefore involves a sequence of processes including information acquisition, interpretation, and behavioral response. Studies show that blind individuals construct spatial knowledge through continuous perceptual experience\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e and that the stability of non-visual environmental cues directly influences spatial confidence and orientation\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eSpatial meaning thus emerges through interpretation rather than through raw sensory input alone. However, many navigation technologies focus primarily on coordinates or object detection without clearly indicating how users should respond to the information provided. When environmental cues are fragmented or inconsistent, interpretive coherence can be disrupted, thereby increasing uncertainty and cognitive load during navigation.\u003c/p\u003e \u003cp\u003eBehavioral Mediation and Travel Intention\u003c/p\u003e \u003cp\u003eMobility decisions are also shaped by psychological processes. The Theory of Planned Behavior explains behavioral intention through attitudes, subjective norms, and perceived behavioral control\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eIn uncertain environments, psychological responses such as anxiety and perceived insecurity may weaken perceived behavioral control and reduce the willingness to travel independently\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e. Empirical studies indicate that environmental uncertainty often influences mobility behavior indirectly through psychological appraisal rather than through physical barriers alone\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eIntegrating psychological variables into mobility analysis therefore helps explain how environmental conditions influence travel intention, particularly for visually impaired users whose mobility experiences are closely related to environmental reliability and perceived safety.\u003c/p\u003e \u003cp\u003eEnvironmental Semantic Reconstruction and Research Hypotheses\u003c/p\u003e \u003cp\u003eAlthough previous research has examined visually impaired mobility through perspectives such as spatial justice, cognitive mapping, and behavioral intention theory, an integrated explanation linking environmental perception, psychological mediation, and behavioral outcomes remains limited\u003csup\u003e\u003cspan additionalcitationids=\"CR27\" citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e. Many studies focus either on accessibility infrastructure or on assistive technologies while treating environmental conditions and behavioral responses as separate domains\u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e,\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eTo address this gap, this study proposes a layered framework of environmental semantic reconstruction. Environmental semantics is defined as the structured organization of environmental meaning that enables individuals to translate perception into action (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThis process operates across three interconnected layers:\u003c/p\u003e \u003cp\u003ePerceptual Semantic Layer: the physical continuity and reliability of environmental cues\u003c/p\u003e \u003cp\u003eInterpretive Semantic Layer: the cognitive integration of spatial information and the evaluation of environmental uncertainty\u003c/p\u003e \u003cp\u003eAction-Oriented Semantic Layer: the translation of interpreted meaning into behavioral control and mobility decisions\u003c/p\u003e \u003cp\u003eEnvironmental semantic reconstruction therefore refers to restoring coherence across these layers when discontinuities arise, ensuring alignment between perception, cognition, and action in complex urban environments.\u003c/p\u003e \u003cp\u003eBased on this theoretical framework, this study proposes that environmental discontinuities generated by fragmented cues increase perceived uncertainty during navigation. Psychological responses, particularly perceived uncertainty and perceived behavioral control, are expected to mediate the relationship between environmental conditions and travel intention. Furthermore, different user groups may adopt distinct mobility strategies when responding to environmental disruptions, resulting in differentiated mobility outcomes.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eResearch Design\u003c/p\u003e\n\u003cp\u003eThis study adopts an empirical research design to investigate the mechanisms influencing mobility behavior among visually impaired pedestrians in urban street environments. The research integrates questionnaire-based data collection with quantitative modeling to examine the relationships between environmental conditions, psychological perception, and travel behavior.\u003c/p\u003e\n\u003cp\u003eThe analytical process consists of four sequential stages: empirical data collection, variable operationalization and preprocessing, quantitative modeling, and structural interpretation of mobility patterns. This stepwise design enables the identification of key influencing factors and heterogeneous mobility profiles within the visually impaired population.\u003c/p\u003e\n\u003cp\u003eThe overall research workflow, from empirical observation to framework development, is illustrated in Fig. \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e\n\u003cp\u003eStudy Population\u003c/p\u003e\n\u003cp\u003eThe study focuses on visually impaired residents in Shanghai. Participants were recruited through collaboration with the Shanghai Association of the Blind and the Jing\u0026rsquo;an District Disabled Persons\u0026rsquo; Federation.\u003c/p\u003e\n\u003cp\u003eData collection took place between December 1 and December 15, 2025, using the Tencent Survey online platform, which supports accessibility features such as screen reader compatibility and voice-assisted navigation. Respondents completed the questionnaire independently using their personal devices.\u003c/p\u003e\n\u003cp\u003eA total of 309 valid responses were obtained. According to statistics from the Shanghai Disabled Persons\u0026rsquo; Federation (2023), the registered visually impaired population in Shanghai is 95,946, meaning the sample represents approximately 0.32% of the population.\u003c/p\u003e\n\u003cp\u003eUser groups were not predefined during data collection. Instead, population heterogeneity was explored during the analytical stage using clustering techniques.\u003c/p\u003e\n\u003cp\u003eParticipation in this study was entirely voluntary. All respondents were fully informed of the study objectives and their right to withdraw at any time prior to completing the questionnaire. This research was conducted in accordance with the Declaration of Helsinki and was approved by the Science and Technology Ethics Committee of Donghua University (Approval Number: RLSSZYJ202603170025). The study was facilitated by the Shanghai and Jing\u0026rsquo;an District Disabled Persons\u0026rsquo; Federations. Written informed consent was obtained from all participants. We further confirm that all participants were adults aged 18 and above, and all responses were anonymized to protect personal identifying information.\u003c/p\u003e\n\u003cp\u003eData Collection\u003c/p\u003e\n\u003cp\u003eThe data were derived from the Shanghai Public Space Experience Questionnaire for Visually Impaired Individuals, designed to capture environmental perception, facility experience, psychological responses, and travel behavior in urban street environments.\u003c/p\u003e\n\u003cp\u003eTo ensure accessibility for participants with different levels of vision, the questionnaire platform supported screen readers and voice-assisted navigation, enabling both blind and low-vision individuals to complete the survey independently.\u003c/p\u003e\n\u003cp\u003eThe questionnaire was structured around five dimensions:\u003c/p\u003e\n\u003cp\u003e(A) Demographic Characteristics and Assistive Devices,\u003c/p\u003e\n\u003cp\u003e(B) Perceived Environmental Barriers,\u003c/p\u003e\n\u003cp\u003e(C/D) Facility Experience,\u003c/p\u003e\n\u003cp\u003e(F) Psychological-Sensory Experience, and\u003c/p\u003e\n\u003cp\u003e(E) Mobility Behavior Intention.\u003c/p\u003e\n\u003cp\u003eMobility intention represents the behavioral outcome in real street contexts, while the remaining dimensions capture environmental and psychological factors that may influence this outcome. A summary of questionnaire variables and their analytical roles is presented in Table \u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\u0026nbsp;\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eOverview of questionnaire dimensions and variable roles.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003eDimension\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eCode\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eItem description\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eScale\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eRole in analysis\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\" colname=\"c1\"\u003e\n \u003cp\u003e\u003cstrong\u003eIndividual attributes\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eA1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eLevel of visual impairment\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eCategorical (Nominal)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eControl variable\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eA2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003ePrimary travel frequency\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eCategorical (Multiple Choice)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eIndependent Variable\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eA3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003ePreferred assistive modes for smart systems\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eCategorical (Multiple Choice)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eDesign Reference\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003e\u003cstrong\u003eEnvironmental barriers\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eB1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eFrequency of physical obstacles (obstruction/broken paving)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eLikert (1\u0026ndash;5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eIndependent Variable\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eB2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eImpact of obstacles on safety and convenience\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eLikert (1\u0026ndash;5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eIndependent Variable\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eB3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eSubjective major difficulties (noise, landmarks, speed)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eCategorical (Multiple Choice)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eIndependent Variable\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003e\u003cstrong\u003eNavigation \u0026amp; Facility Experience\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eC1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eConvenience of facility usage in public spaces\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eLikert (1\u0026ndash;5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eMediating Variable\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eC2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eDifficulty in indoor spatial orientation and identification\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eLikert (1\u0026ndash;5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eMediating Variable\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eD1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eDependence on physical and digital navigation aids\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eLikert (1\u0026ndash;5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eIndependent Variable\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eD2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eOverall satisfaction with current assistive methods\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eCategorical (Multiple Choice)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eOutcome Variable / Feature\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003e\u003cstrong\u003eTravel Intention \u0026amp; Behavior\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eE1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eWillingness to reduce or avoid outdoor travel\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eCategorical (Ordinal)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eDependent Variable\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eE2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003ePrimary reasons for travel avoidance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eCategorical (Multiple Choice)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eDependent Variable\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003e\u003cstrong\u003ePsychological Perception\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eF1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003ePsychological anxiety in unfamiliar environments\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eLikert (1\u0026ndash;5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eMediating Variable\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eF2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003ePerceived lack of safety during travel\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eLikert (1\u0026ndash;5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eMediating Variable\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eF3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eSpecific psychological stressors/worries (e.g., collisions, getting lost)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eCategorical (Multiple Choice)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eFeature Engineering / Explainability\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colname=\"c1\"\u003e\n \u003cp\u003e\u003cstrong\u003eQualitative Support\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c2\"\u003e\n \u003cp\u003eG1/G2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c3\"\u003e\n \u003cp\u003eOverall satisfaction and open-ended suggestions\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c4\"\u003e\n \u003cp\u003eLikert / Qualitative\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\" colname=\"c5\"\u003e\n \u003cp\u003eQualitative Support\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\u003eDuring the distribution process, on-site briefing sessions were also conducted to explain the questionnaire and assist participants when necessary.\u003c/p\u003e\n\u003cp\u003eVariable Operationalization and Data Preprocessing\u003c/p\u003e\n\u003cp\u003eQuestionnaire responses\u0026mdash;primarily reflecting subjective perceptions, environmental experiences, and behavioral tendencies\u0026mdash;were transformed into quantifiable variables suitable for statistical and machine learning analysis. During this process, each variable was operationalized to retain its conceptual meaning while ensuring numerical interpretability and analytical consistency.\u003c/p\u003e\n\u003cp\u003eBased on the questionnaire structure, variables were organized into five conceptual dimensions: demographic characteristics and assistive device usage, perceived environmental barriers, facility and navigation experience, psychological perception, and travel behavior intention. These dimensions correspond to the analytical framework developed in this study and provide the basis for subsequent modeling.\u003c/p\u003e\n\u003cp\u003ePrior to analysis, the raw survey data underwent systematic data cleaning and consistency verification. Responses were screened for logical consistency and valid response ranges, and incomplete or inconsistent entries were removed. Variables that were redundant, weakly related to the research objectives, or highly collinear variables were excluded during the feature screening stage to mitigate multicollinearity. After this process, a total of fifteen core features were retained as the primary explanatory variables for modeling.\u003c/p\u003e\n\u003cp\u003eTo prepare the dataset for machine learning analysis, several preprocessing procedures were applied. Continuous and ordinal variables were standardized using z-score normalization to ensure comparability across different measurement scales. Categorical variables were encoded as binary or dummy variables while preserving ordinal structure where applicable. All preprocessing and modeling procedures were implemented in Python (version 3.12).\u003c/p\u003e\n\u003cp\u003eAfter preprocessing, the resulting dataset contained no missing values and was structured as a feature matrix suitable for machine learning modeling. For supervised learning analysis, the travel intention variable\u0026mdash;specifically whether respondents would avoid certain street environments\u0026mdash;was binarized as the target variable, while all remaining variables were used as input features.\u003c/p\u003e\n\u003cp\u003eAnalytical Strategy\u003c/p\u003e\n\u003cp\u003eUsing the structured dataset described above, this study developed a problem-oriented analytical framework to examine travel avoidance among visually impaired pedestrians.\u003c/p\u003e\n\u003cp\u003eThe objective extends beyond predictive accuracy and focuses on identifying structurally significant factors and population heterogeneity. Accordingly, the analysis combines supervised classification with unsupervised clustering.\u003c/p\u003e\n\u003cp\u003eThe analytical procedure is summarized in Fig. \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e\n\u003cp\u003eFirst, supervised classification models were used to examine relationships between environmental and psychological variables and travel avoidance behavior. Three representative models were compared:\u003c/p\u003e\n\u003cp\u003eLogistic regression, which serves as a baseline model providing interpretable coefficients indicating the direction and magnitude of influence.\u003c/p\u003e\n\u003cp\u003eRandom forest (RF), an ensemble tree-based model capable of capturing nonlinear relationships and variable interactions while providing feature importance measures.\u003c/p\u003e\n\u003cp\u003eGradient boosting decision tree (GBDT), an iterative ensemble method that improves predictive performance by sequentially modeling residual errors.\u003c/p\u003e\n\u003cp\u003eTo ensure the generalizability of the results, all models were evaluated using k-fold cross-validation, mitigating the dependency on a single train-test split. Regularization techniques such as parameter tuning and tree-depth limitation were applied to reduce overfitting.\u003c/p\u003e\n\u003cp\u003eModel performance was assessed using accuracy and F1-score, with the latter emphasized due to potential class imbalance. To ensure predictive reliability, random forest feature importance was derived using Gini impurity to quantify the contribution of environmental and psychological factors. For population heterogeneity, k-means clustering was optimized via the Elbow Method and Silhouette Coefficient to determine the most representative mobility profiles.\u003c/p\u003e\n\u003cp\u003eFeature importance derived from the random forest model was further analyzed to identify variables with strong explanatory value.\u003c/p\u003e\n\u003cp\u003eTo explore heterogeneity within the sample, clustering analysis was conducted using key perceptual and behavioral variables, including environmental barrier perception, psychological burden, and assistive technology use. This analysis revealed distinct mobility profiles characterized by different combinations of perception, anxiety, and technology reliance.\u003c/p\u003e\n\u003cp\u003eThe integration of classification and clustering enables the identification of both key influencing factors and differentiated mobility patterns within the visually impaired population, providing the empirical foundation for the structural interpretation presented in the Results section.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eBased on the analytical framework and model specifications outlined above, this section presents the empirical results of the urban street mobility analysis for visually impaired individuals. The findings are organized into three components: model performance in predicting travel intention, identification of key influencing factors, and group differences within the sample. All analyses were conducted on the processed dataset following variable operationalization and preprocessing.\u003c/p\u003e \u003cp\u003ePerformance Evaluation of Travel Intention Prediction Models\u003c/p\u003e \u003cp\u003eAfter data preprocessing and feature engineering, three models\u0026mdash;logistic regression, random forest, and gradient boosting decision tree (GBDT)\u0026mdash;were implemented to predict the binary outcome of whether individuals tend to reduce outdoor travel. Model performance was evaluated using five-fold cross-validation.\u003c/p\u003e \u003cp\u003eAs shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e4\u003c/span\u003e, the three models differ in predictive performance. Random forest achieves the highest overall accuracy (approximately 78%) and an F1-score close to 0.78, demonstrating balanced precision and recall in identifying individuals who reduce outdoor mobility. In contrast, logistic regression performs less effectively, suggesting limited ability to capture nonlinear relationships within the dataset. The GBDT model reaches accuracy levels comparable to random forest; however, its performance fluctuates more across validation folds, indicating slightly lower stability.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eAcross all evaluation metrics, random forest consistently outperforms logistic regression and GBDT. Random forest was therefore selected as the primary model for subsequent feature importance analysis due to its superior predictive performance\u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eFeature Importance Analysis of Influencing Factors\u003c/p\u003e \u003cp\u003eFollowing model selection, feature importance scores derived from the random forest model were examined to identify variables most strongly associated with reduced outdoor mobility (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e5\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe results indicate that psychological anxiety ranks highest among all predictors, highlighting its central role in travel avoidance behavior. This finding suggests that mobility decisions are shaped not only by objective environmental conditions but also by perceived uncertainty during movement. When environmental feedback lacks reliability and predictability, perceived safety diminishes, increasing the likelihood of avoiding independent travel.\u003c/p\u003e \u003cp\u003eIn addition to psychological anxiety, the frequency of occupied tactile paving and wayfinding difficulty also exhibit high importance scores. Occupied tactile paving reflects disruptions in environmental continuity, whereas wayfinding difficulty points to informational gaps when locating entrances, crossings, or key spatial nodes. Together, these variables suggest a pathway through which environmental disruptions elevate psychological stress, ultimately reducing willingness to travel.\u003c/p\u003e \u003cp\u003eThese results highlight the joint influence of environmental conditions and psychological perception on travel intention.\u003c/p\u003e \u003cp\u003eUser Profile Analysis Based on k-means Clustering\u003c/p\u003e \u003cp\u003eBeyond identifying overall influencing factors, the study further explores heterogeneity within the visually impaired population through clustering analysis. Individuals were grouped using k-means clustering based on perceptual and behavioral characteristics.\u003c/p\u003e \u003cp\u003eClustering variables were selected according to both random forest feature importance results and theoretical considerations. Specifically, visual impairment level (A1), assistive device usage (A2), perceived environmental barriers (B1), and psychological anxiety (F1) were included as input features. Collectively, these variables capture environmental exposure, psychological state, and coping strategies during urban street travel. The optimal number of clusters was determined as K\u0026thinsp;=\u0026thinsp;4 using the elbow method and silhouette coefficient evaluation. The clustering solution was visualized using principal component analysis (PCA), as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e6\u003c/span\u003e. The four clusters display distinct configurations in psychological perception, environmental experience, and assistive strategy use.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eStable Traditional Users (Cluster 1)\u003c/p\u003e \u003cp\u003eCluster 1 reports the lowest level of psychological anxiety (\u003cem\u003eM\u003c/em\u003e\u0026thinsp;=\u0026thinsp;2.54, \u003cem\u003eSD\u003c/em\u003e\u0026thinsp;=\u0026thinsp;1.07) and minimal reliance on smartphone-based navigation (\u003cem\u003eM\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.01, \u003cem\u003eSD\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.11). Perceived environmental barriers are also relatively low compared with other clusters.\u003c/p\u003e \u003cp\u003eMembers of this group primarily depend on traditional assistive strategies, such as white canes and accumulated spatial familiarity. The low anxiety level suggests well-developed coping mechanisms and adaptive strategies formed through long-term environmental experience. Rather than relying on digital systems, their mobility patterns reflect established spatial knowledge and confidence in conventional tactile and spatial cues. This pattern suggests that mobility in this group relies primarily on environmental familiarity rather than technological assistance.\u003c/p\u003e \u003cp\u003eVulnerable Environment-Dependent Users (Cluster 2)\u003c/p\u003e \u003cp\u003eCluster 2 exhibits the highest level of psychological anxiety (\u003cem\u003eM\u003c/em\u003e\u0026thinsp;=\u0026thinsp;3.53, \u003cem\u003eSD\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.98) and the strongest perception of environmental barriers, including tactile paving obstruction and wayfinding difficulty, indicating that their elevated anxiety is closely coupled with frequent encounters with physical environmental disruptions.\u003c/p\u003e \u003cp\u003eDespite facing substantial environmental challenges, smartphone usage remains relatively low (\u003cem\u003eM\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.18, \u003cem\u003eSD\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.38), indicating limited technological compensation. The combination of high environmental exposure and low technological reliance characterizes a vulnerable mobility profile. Travel behavior in this group appears strongly constrained by environmental uncertainty, with relatively low reliance on technological assistance.\u003c/p\u003e \u003cp\u003eTechnology-Assisted but Environmentally Sensitive Users (Cluster 3)\u003c/p\u003e \u003cp\u003eCluster 3 shows elevated psychological anxiety (\u003cem\u003eM\u003c/em\u003e\u0026thinsp;=\u0026thinsp;3.40, \u003cem\u003eSD\u003c/em\u003e\u0026thinsp;=\u0026thinsp;1.31) and high perception of environmental barriers, while demonstrating moderate adoption of smartphone-based navigation (\u003cem\u003eM\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.53, \u003cem\u003eSD\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.50).\u003c/p\u003e \u003cp\u003eAlthough assistive technology is actively used, environmental uncertainty continues to affect psychological stability and mobility confidence, suggestive of a persistent gap between digital information and physical reality. This indicates that while smartphone-based navigation provides supplementary information, it cannot fully mitigate the mobility risks induced by severe physical infrastructure discontinuities.\u003c/p\u003e \u003cp\u003eTechnology-Empowered Low-Anxiety Users (Cluster 4)\u003c/p\u003e \u003cp\u003eCluster 4 exhibits the highest smartphone usage rate (\u003cem\u003eM\u003c/em\u003e\u0026thinsp;=\u0026thinsp;1.00, \u003cem\u003eSD\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.00) and comparatively low levels of anxiety (\u003cem\u003eM\u003c/em\u003e\u0026thinsp;=\u0026thinsp;2.53, \u003cem\u003eSD\u003c/em\u003e\u0026thinsp;=\u0026thinsp;1.02).\u003c/p\u003e \u003cp\u003eThis mobility profile suggests strong technological adaptation and effective use of assistive navigation tools. In this group, technology appears to function as a stabilizing mechanism that reduces uncertainty and supports independent mobility.\u003c/p\u003e \u003cp\u003eStatistical Validation of Cluster Differences\u003c/p\u003e \u003cp\u003eTo evaluate whether the four-cluster solution represents substantively distinct mobility profiles, a one-way analysis of variance (ANOVA) was conducted with cluster membership (K\u0026thinsp;=\u0026thinsp;4) as the grouping variable. Five variables were selected for validation based on theoretical relevance and feature importance results: psychological anxiety, frequency of tactile paving obstruction, frequency of tactile paving defects, frequency of wayfinding difficulty, and smartphone usage rate for navigation.\u003c/p\u003e \u003cp\u003eAs shown in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, significant differences were observed across clusters for all variables (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Effect sizes (\u003cem\u003eη\u0026sup2;\u003c/em\u003e) ranged from 0.159 to 0.646, indicating moderate to large between-group effects. Smartphone usage rate demonstrated the largest effect size (\u003cem\u003eη\u0026sup2;\u003c/em\u003e = 0.646), underscoring the structurally differentiating role of technological adaptation across groups. Environmental barrier indicators and psychological anxiety also showed substantial effect sizes, indicating clear differences among the clusters.\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\u003eOne-way ANOVA results across the four mobility clusters.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"8\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" 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=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCluster 1 mean\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCluster 2 mean\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eCluster 3 mean\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eCluster 4 mean\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e\u003cem\u003eF value\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003e\u003cem\u003ep value\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u003cem\u003eη\u0026sup2;\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLevel of psychological anxiety\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e2.54\u0026thinsp;\u0026plusmn;\u0026thinsp;1.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e3.53\u0026thinsp;\u0026plusmn;\u0026thinsp;0.98\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e \u003cp\u003e3.40\u0026thinsp;\u0026plusmn;\u0026thinsp;1.31\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c5\"\u003e \u003cp\u003e2.53\u0026thinsp;\u0026plusmn;\u0026thinsp;1.02\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e19.169\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.159\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFrequency of tactile paving obstruction\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e2.95\u0026thinsp;\u0026plusmn;\u0026thinsp;0.90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e4.27\u0026thinsp;\u0026plusmn;\u0026thinsp;0.61\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e \u003cp\u003e3.97\u0026thinsp;\u0026plusmn;\u0026thinsp;0.86\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c5\"\u003e \u003cp\u003e3.44\u0026thinsp;\u0026plusmn;\u0026thinsp;0.85\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e40.387\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.284\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFrequency of tactile paving defects\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e2.35\u0026thinsp;\u0026plusmn;\u0026thinsp;0.74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e3.72\u0026thinsp;\u0026plusmn;\u0026thinsp;0.77\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e \u003cp\u003e3.50\u0026thinsp;\u0026plusmn;\u0026thinsp;0.98\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c5\"\u003e \u003cp\u003e2.84\u0026thinsp;\u0026plusmn;\u0026thinsp;0.95\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e41.524\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.29\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFrequency of difficulty finding one's way\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e2.28\u0026thinsp;\u0026plusmn;\u0026thinsp;0.79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e3.63\u0026thinsp;\u0026plusmn;\u0026thinsp;0.79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e \u003cp\u003e3.55\u0026thinsp;\u0026plusmn;\u0026thinsp;0.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c5\"\u003e \u003cp\u003e2.82\u0026thinsp;\u0026plusmn;\u0026thinsp;0.87\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e48.891\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.325\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSmartphone usage rate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e0.01\u0026thinsp;\u0026plusmn;\u0026thinsp;0.11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e0.18\u0026thinsp;\u0026plusmn;\u0026thinsp;0.38\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e \u003cp\u003e0.53\u0026thinsp;\u0026plusmn;\u0026thinsp;0.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c5\"\u003e \u003cp\u003e1.00\u0026thinsp;\u0026plusmn;\u0026thinsp;0.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e185.238\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.646\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\u003eTukey\u0026rsquo;s post hoc tests (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e) confirmed these distinct patterns: Clusters 2 and 3 reported significantly higher environmental barriers and psychological anxiety than Clusters 1 and 4 (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001), while Cluster 4 showed the highest smartphone usage.\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 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eTukey post hoc comparisons of significant differences across the four mobility clusters.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\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=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eComparison\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMean Difference\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003ep value\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e95% CI\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003eLevel of psychological anxiety\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC1 vs C2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.99\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[0.548, 1.431]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC1 vs C3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.854\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[0.374, 1.335]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC2 vs C4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;1.004\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[\u0026minus;\u0026thinsp;1.438, \u0026minus;\u0026thinsp;0.570]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC3 vs C4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;0.869\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[\u0026minus;\u0026thinsp;1.342, \u0026minus;\u0026thinsp;0.395]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003eFrequency of tactile paving obstruction\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC1 vs C2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.314\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[0.984, 1.644]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC1 vs C4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.486\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[0.166, 0.806]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC2 vs C4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;0.828\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[\u0026minus;\u0026thinsp;1.152, \u0026minus;\u0026thinsp;0.503]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC3 vs C4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;0.527\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[\u0026minus;\u0026thinsp;0.881, \u0026minus;\u0026thinsp;0.173]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003eFrequency of tactile paving defects\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC1 vs C2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.372\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[1.023, 1.721]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC1 vs C4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.493\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[0.155, 0.832]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC2 vs C4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;0.879\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[\u0026minus;\u0026thinsp;1.222, \u0026minus;\u0026thinsp;0.536]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC3 vs C4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;0.657\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[\u0026minus;\u0026thinsp;1.032, \u0026minus;\u0026thinsp;0.283]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003eFrequency of wayfinding difficulty\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC1 vs C2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.356\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[1.028, 1.683]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC1 vs C4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.543\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[0.225, 0.861]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC2 vs C4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;0.813\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[\u0026minus;\u0026thinsp;1.135, \u0026minus;\u0026thinsp;0.491]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC3 vs C4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026minus;0.732\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[\u0026minus;\u0026thinsp;1.083, \u0026minus;\u0026thinsp;0.380]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"5\" rowspan=\"6\"\u003e \u003cp\u003eSmartphone usage rate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC1 vs C2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.165\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[0.044, 0.286]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC1 vs C3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.522\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[0.391, 0.654]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC1 vs C4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.988\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[0.871, 1.105]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC2 vs C3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.357\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[0.225, 0.490]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC2 vs C4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.823\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[0.704, 0.942]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eC3 vs C4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.466\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[0.336, 0.595]\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\u003eThese differences are visually summarized in Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e7\u003c/span\u003e, which presents the mean values of the five validation variables across clusters. Clusters 2 and 3 form a high-barrier, high-anxiety profile, while Cluster 4 is characterized by strong technological adaptation despite moderate environmental exposure. Cluster 1 represents a relatively low-barrier, low-anxiety profile with limited reliance on digital assistance.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eTaken together, the ANOVA and post hoc analyses confirm that the clustering solution reflects statistically distinguishable mobility profiles rather than arbitrary algorithmic segmentation. The integration of clustering and statistical validation strengthens confidence that the identified profiles represent structurally differentiated mobility patterns. The four groups differ systematically in perceived environmental barriers, psychological response, and assistive behavior, providing a robust empirical basis for subsequent strategy differentiation.\u003c/p\u003e \u003cp\u003eWhile clustering identifies statistically distinct behavioral configurations, the observed differences also reflect variations in environmental perception and psychological response among user groups. These empirical patterns provide the basis for the environmental semantic reconstruction framework introduced in the following section.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eFrom Behavioral Clusters to Differences in Environmental Semantics\u003c/p\u003e\n\u003cp\u003eThis study demonstrates that mobility stability depends on the continuity of environmental meaning\u0026mdash;perceived intelligibility and predictability\u0026mdash;rather than mere infrastructure presence. Unlike conventional research treating accessibility as physical provision, our findings reveal that mobility relies on whether information can be reliably perceived, cognitively integrated, and translated into action. This aligns with research emphasizing perceived accessibility in shaping behavior\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e,\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eThe four statistically distinct clusters identified in the Results reflect not merely behavioral variation, but differentiated pathways through which environmental meaning is constructed and stabilized.\u003c/p\u003e\n\u003cp\u003eImportantly, these clusters do not represent differences in capability. They reflect differences in semantic interpretation pathways. Urban streets function not as neutral information carriers, but as environments whose meaning varies depending on cognitive habits and technological engagement.\u003c/p\u003e\n\u003cp\u003eAlthough the analysis primarily relies on perception-based questionnaire data, many of the measured variables correspond to observable environmental conditions, such as tactile paving obstruction, facility defects, and navigation reliability. The findings therefore reflect the interaction between objective environmental cues and subjective interpretation, rather than subjective perception alone.\u003c/p\u003e\n\u003cp\u003eLinking Behavioral Clusters to Environmental Reconstruction Strategies\u003c/p\u003e\n\u003cp\u003eBehavioral clusters reveal distinct semantic disruptions manifesting from infrastructure failures. According to Lynch\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e, legibility is essential for navigation; for visually impaired pedestrians, this relies on consistent tactile paving and auditory signals. To bridge behaviors with infrastructure, we categorize elements by functional role: tactile paving and curb ramps form the perceptual-layer (primary signals), while auditory signals and beacons constitute the interpretive-layer (supplementary information).\u003c/p\u003e\n\u003cp\u003eThe four mobility clusters identified in this study can therefore be interpreted as reflecting different patterns of environmental cue disruption (as quantified in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e):\u003c/p\u003e\n\u003cp\u003eCluster 1 (Stable Traditional Users) exhibits relatively low perceived environmental barriers and anxiety, indicating that these individuals can still rely on the existing \u0026ldquo;legibility\u0026rdquo; of the environment. Their mobility suggests that current street infrastructure supports navigation when tactile paths remain continuous and predictable. For this group, environmental reconstruction involves preventive maintenance to ensure that the existing semantic structure does not degrade.\u003c/p\u003e\n\u003cp\u003eCluster 2 (Vulnerable Environment-Dependent Users) represents a perceptual-layer disruption. The random forest analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e5\u003c/span\u003e) and the significantly higher mean scores for tactile paving issues in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e provide empirical evidence for this: the high importance scores for \u0026quot;tactile paving obstruction\u0026quot; and \u0026ldquo;defects\u0026rdquo; demonstrate that when these specific physical objects fail, the entire perceptual system of the pedestrian collapses, confirming that physical defects are the primary external triggers for the top-ranked psychological anxiety (Mean\u0026thinsp;=\u0026thinsp;3.53, Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e) observed in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e5\u003c/span\u003e. This statistical coupling between physical object failure, psychological response, and behavioral outcome validates the identification of these facilities as critical perceptual-layer nodes. Their high anxiety reflects a physical breakdown in the environment\u0026rsquo;s primary communication system. When the tactile \u0026ldquo;language\u0026rdquo; of the street is interrupted by obstacles, the environment becomes semantically \u0026ldquo;silent\u0026rdquo; or \u0026ldquo;misleading.\u0026rdquo; Consequently, stabilizing tactile information through systematic maintenance is the most direct intervention to restore environmental predictability.\u003c/p\u003e\n\u003cp\u003eCluster 3 (Technology-Assisted but Environmentally Sensitive Users) highlights an interpretive-layer mismatch. This pattern suggests that while digital navigation tools provide supplementary information, they often fail to align with physical environmental cues. According to James J. Gibson\u0026rsquo;s\u003csup\u003e34\u003c/sup\u003e ecological perception framework, navigation emerges from environmental affordances. If digital guidance does not correspond to physical tactile paving or crossing signals, the \u0026ldquo;information mismatch\u0026rdquo; undermines confidence. Reconstruction for this group must focus on the semantic integration of digital and physical layers. Specifically, this involves deploying Bluetooth beacons or RFID markers that trigger real-time auditory descriptions of intersection geometries to ensure they match the existing physical tactile cues.\u003c/p\u003e\n\u003cp\u003eCluster 4 (Technology-Empowered Low-Anxiety Users) demonstrates the potential of multi-modal environmental information. Their lower anxiety levels suggest that a redundant information system (physical\u0026thinsp;+\u0026thinsp;digital) can compensate for certain environmental limitations. For these users, reconstruction involves enhancing the coordination between physical infrastructure and digital data to create a seamless, multi-layered environmental semantic network.Such a network requires the high-precision coordination between physical sidewalk integrity and dynamic digital data, ensuring that real-time hazard alerts are seamlessly integrated with the stable infrastructure base.\u003c/p\u003e\n\u003cp\u003eIn summary, these behavioral clusters serve as diagnostic indicators of environmental performance. Rather than simply representing user preferences, they reveal where and how the urban street fabric fails to provide the necessary semantic cues for stable mobility.\u003c/p\u003e\n\u003cp\u003eTheoretical and Mechanism Verification\u003c/p\u003e\n\u003cp\u003eAlthough lacking VR testing, the identified links between cue disruption and anxiety align with Lynch\u0026rsquo;s\u003csup\u003e4\u003c/sup\u003e legibility and Gibson\u0026rsquo;s\u003csup\u003e34\u003c/sup\u003e affordance theories. This convergence of machine learning evidence and spatial theory validates the hierarchical framework, transforming pedestrian feedback into objective performance indicators. The convergence of machine learning evidence and spatial perception theory provides a robust basis for the proposed hierarchical reconstruction. This approach aligns with recent calls for hierarchical strategies in tactile paving deployment\u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e and mirrors the broader transition in inclusive governance from physical provision toward technologically-integrated information reliability\u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e. By transforming subjective pedestrian feedback into objective infrastructure performance indicators, this framework acts as a structural interpretation of established mechanisms.\u003c/p\u003e\n\u003cp\u003eA Multihierarchical Environmental Semantic Reconstruction Framework\u003c/p\u003e\n\u003cp\u003eBuilding on these differentiated patterns, environmental semantic reconstruction can be conceptualized as a multihierarchical framework. Mobility stability emerges from continuity across three interdependent layers: perceptual acquisition, interpretive integration, and action translation.\u003c/p\u003e\n\u003cp\u003eThe Perceptual Layer concerns the acquisition of environmental cues, including tactile continuity and spatial structure. Instability at this level initiates semantic breakdown.\u003c/p\u003e\n\u003cp\u003eThe Interpretive Layer involves cognitive integration and uncertainty evaluation. Disruption manifests as anxiety and reduced confidence, even where partial cues are present.\u003c/p\u003e\n\u003cp\u003eThe Action-Oriented Layer translates interpreted information into executable movement. Technological systems often intervene here through route confirmation and real-time adjustment feedback.\u003c/p\u003e\n\u003cp\u003eMobility stability is therefore not a direct function of infrastructure provision or technological presence alone. It depends on whether semantic continuity is maintained\u0026mdash;or reconstructed\u0026mdash;across these layers.\u003c/p\u003e\n\u003cp\u003eThe decisive variable across clusters is therefore not the intensity of technological assistance, but the layer at which reconstruction intervenes. Early-layer stabilization generates organic continuity; later-layer compensation produces mediated stability. Where no effective intervention exists, instability propagates.\u003c/p\u003e\n\u003cp\u003eEnvironmental semantic reconstruction is thus defined as the structural reorganization of environmental meaning to restore coherence between perception, cognition, and action.\u003c/p\u003e\n\u003cp\u003eOperationally, reconstruction involves:\u003c/p\u003e\n\u003cp\u003e1. Restoring perceptual continuity in street infrastructure;\u003c/p\u003e\n\u003cp\u003e2. Enhancing interpretive predictability in spatial cognition;\u003c/p\u003e\n\u003cp\u003e3. Reinforcing action-level confirmation mechanisms.\u003c/p\u003e\n\u003cp\u003eThis layered model clarifies how differentiated mobility profiles correspond to distinct interruption points and intervention priorities (Fig. \u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e8\u003c/span\u003e, Fig. \u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e9\u003c/span\u003e).\u003c/p\u003e\n\u003cp\u003eFrom infrastructure provision to semantic continuity in accessibility governance\u003c/p\u003e\n\u003cp\u003eGovernance must shift from facility expansion to ensuring semantic reliability and synchronization. At the perceptual level, metrics should prioritize obstruction frequency and maintenance response over mere installation rates, institutionalizing enforcement against sidewalk occupation. At the interpretive level, inter-departmental coordination is essential to synchronize infrastructure upgrades and reduce environmental fragmentation. At the action level, digital systems should serve as confirmation mechanisms; thus, data accuracy and synchronization with physical conditions are critical.\u003c/p\u003e\n\u003cp\u003eDifferentiated cluster needs demand targeted priorities: environmentally vulnerable users require perceptual stabilization, while technology-dependent groups need digital-physical integration and data reliability. Moving beyond the \u0026ldquo;average user\u0026rdquo; assumption, reframing accessibility as semantic stability promotes a continuity-oriented model of inclusive street management. By empirically linking perceptual uncertainty with behavioral adaptation, this study provides a framework for data-informed, inclusive urban governance.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eAlthough the empirical data were collected in Shanghai, the environmental semantic reconstruction framework proposed in this study is conceptually transferable to other high-density urban contexts facing comparable accessibility challenges. By integrating random forest modeling, k-means clustering, and statistical validation, this study provides a data-driven examination of mobility intentions among visually impaired pedestrians. The findings indicate that mobility stability in urban street environments is not determined solely by the mere provision of accessibility infrastructure. Rather, it emerges from the continuity\u0026mdash;or breakdown\u0026mdash;of environmental meaning across perceptual, interpretive, and action-oriented layers of navigation.\u003c/p\u003e \u003cp\u003eEven in cities where tactile paving systems and barrier-free facilities are formally implemented, discontinuous environmental cues and unreliable infrastructure performance can generate psychological anxiety and erode mobility confidence. These findings suggest that accessibility research and governance have historically prioritized physical installation while insufficiently addressing the semiotic and cognitive dimensions of the lived environmental experience.\u003c/p\u003e \u003cp\u003eBy identifying four differentiated user clusters, this study demonstrates that mobility stability develops through distinct environmental disruption-behavioral compensation pathways. For some individuals, accumulated environmental familiarity sustains perceptual continuity. For others, technological mediation at the action layer partially or fully compensates for perceptual disruption. Mobility outcomes are therefore shaped not by the intensity of technological deployment per se, but by the level within the perceptual-interpretive-action chain at which semantic reconstruction intervenes.\u003c/p\u003e \u003cp\u003eFrom a governance perspective, the results highlight the need to complement infrastructure provision with reliability-oriented management of accessibility systems. Rather than focusing solely on the installation of barrier-free facilities, governance frameworks should consider the everyday usability and environmental consistency of accessibility infrastructure as central performance indicators.\u003c/p\u003e \u003cp\u003eWithin this perspective, the environmental semantic reconstruction framework provides a conceptual basis for integrating physical design, digital sensing, and adaptive governance mechanisms. Reconstruction therefore involves not merely increasing the amount of environmental information, but ensuring that environmental cues remain interpretable and dependable throughout the perceptual-interpretive-action chain. Emerging smart-city technologies, such as embedded sensing systems for monitoring tactile paving conditions and obstruction patterns, may support this objective by enabling data-informed maintenance and responsive management of urban accessibility infrastructure\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e,\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e. Importantly, such technological interventions should be aligned with differentiated user profiles so that reconstruction efforts address the specific layer at which semantic disruption occurs.\u003c/p\u003e \u003cp\u003eSeveral limitations warrant consideration. The study is constrained by its sampling scope and single-city context. Future research could incorporate longitudinal field validation, cross-city comparative analysis, and prototype implementation testing to further examine how layered semantic reconstruction influences long-term behavioral stability and psychological adaptation. A pilot smart tactile paving system currently under development in Shanghai seeks to operationalize the framework proposed here. Systematic evaluation of this deployment will provide empirical evidence for refining the interaction among environmental reliability, technological mediation, and user experience.\u003c/p\u003e \u003cp\u003eMore broadly, the framework may be extended to other vulnerable populations, including older adults and individuals with mobility impairments, to assess its applicability within wider inclusive urban governance contexts.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eFunding\u003c/p\u003e\n\u003cp\u003eThis research was supported by the Shanghai Philosophy and Social Sciences Planning Project (Grant No. 2024BCK012), funded by the Shanghai Philosophy and Social Sciences Planning Office (Project title: \u0026ldquo;Research on the Spatial Agglomeration and Symbiotic Development Model of Shanghai\u0026rsquo;s Digital Creative Industry\u0026rdquo;).\u003c/p\u003e\n\u003cp\u003eAuthor contributions\u003c/p\u003e\n\u003cp\u003eS.S. and Y.C. contributed equally to this work.\u003c/p\u003e\n\u003cp\u003eS.S. conceived and designed the research and supervised the study.\u003c/p\u003e\n\u003cp\u003eY.C. performed the data analysis and drafted the manuscript.\u003c/p\u003e\n\u003cp\u003eJ.Z. supervised the research and revised the manuscript.\u003c/p\u003e\n\u003cp\u003eAll authors reviewed and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003eCompeting interests\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003eData availability\u003c/p\u003e\n\u003cp\u003eThe datasets generated and analyzed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003eEthics approval and consent to participate\u003c/p\u003e\n\u003cp\u003eThis study was conducted in accordance with the Declaration of Helsinki. The research protocol and questionnaire were reviewed and approved by the Science and Technology Ethics Committee of Donghua University (Approval Number: RLSSZYJ202603170025), with additional administrative support and review by the Shanghai Disabled Persons\u0026rsquo; Federation and the Jing\u0026rsquo;an District Disabled Persons\u0026rsquo; Federation. Written informed consent was obtained from all participants prior to their involvement. We confirm that all participants were adults aged 18 and above; no minors were involved in this research. All data were fully anonymized before analysis to ensure privacy.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eWorld Health Organization. \u003cem\u003eWorld Report on Vision\u003c/em\u003e (Geneva, 2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGBD 2020 Blindness and Vision Impairment Collaborators. Causes of blindness and vision impairment in 2020 and trends over 30 years, and prevalence of avoidable blindness in relation to VISION 2020: the Right to Sight: an analysis for the Global Burden of Disease Study. \u003cem\u003eLancet Global Health\u003c/em\u003e. \u003cb\u003e9\u003c/b\u003e, e144\u0026ndash;e160 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZou, M. et al. Prevalence of visual impairment among older Chinese population: A systematic review and meta-analysis. \u003cem\u003eJ. Glob Health\u003c/em\u003e. \u003cb\u003e11\u003c/b\u003e, 08004 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKevin Lynch. \u003cem\u003eThe Image of the City\u003c/em\u003e (The MIT, 1960).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePassini, R. Wayfinding design: logic, application and some thoughts on universality. \u003cem\u003eDes. Stud.\u003c/em\u003e \u003cb\u003e17\u003c/b\u003e, 319\u0026ndash;331 (1996).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHersh, M. \u0026amp; Deafblind People Communication, Independence, and Isolation. \u003cem\u003eJ Deaf Stud. Deaf Educ.\u003c/em\u003e \u003cb\u003e18\u003c/b\u003e, 446\u0026ndash;463 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGori, M., Cappagli, G., Baud-Bovy, G. \u0026amp; Finocchietti, S. Shape Perception and Navigation in Blind Adults. \u003cem\u003eFront Psychol\u003c/em\u003e \u003cb\u003e8\u003c/b\u003e, (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu, Y., Li, H. \u0026amp; Pan, Y. Understanding perceived ride safety and trust formation in robotaxi services under day and night conditions. \u003cem\u003eSCIENTIFIC REPORTS\u003c/em\u003e \u003cb\u003e15\u003c/b\u003e (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUN-Habitat. The Value of Sustainable Urbanization. (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchinazi, V. R., Thrash, T. \u0026amp; Chebat, D. Spatial navigation by congenitally blind individuals. \u003cem\u003eWires Cogn. Sci.\u003c/em\u003e \u003cb\u003e7\u003c/b\u003e, 37\u0026ndash;58 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang, Z., Sun, T., Fisher, T. \u0026amp; Wang, H. The relationships between the campus built environment and walking activity. \u003cem\u003eSCIENTIFIC REPORTS\u003c/em\u003e \u003cb\u003e14\u003c/b\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou, K., Hu, C., Zhang, H., Hu, Y. \u0026amp; Xie, B. Why do we hardly see people with visual impairments in the street? A case study of Changsha, China. \u003cem\u003eAppl. Geogr.\u003c/em\u003e \u003cb\u003e110\u003c/b\u003e, 102043 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRan, L., Helal, S. \u0026amp; Moore, S. Drishti: an integrated indoor/outdoor blind navigation system and service. in \u003cem\u003eSecond IEEE Annual Conference on Pervasive Computing and Communications\u003c/em\u003e, Proceedings of the 23\u0026ndash;30 (IEEE, Orlando, FL, USA, 2004). (2004). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1109/PERCOM.2004.1276842\u003c/span\u003e\u003cspan address=\"10.1109/PERCOM.2004.1276842\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMascetti, S., Ahmetovic, D., Gerino, A., Bernareggi, C. \u0026amp; ZebraRecognizer Pedestrian crossing recognition for people with visual impairment or blindness. \u003cem\u003ePattern Recogn.\u003c/em\u003e \u003cb\u003e60\u003c/b\u003e, 405\u0026ndash;419 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBhowmick, A. \u0026amp; Hazarika, S. M. An insight into assistive technology for the visually impaired and blind people: state-of-the-art and future trends. \u003cem\u003eJ. Multimodal User Interfaces\u003c/em\u003e. \u003cb\u003e11\u003c/b\u003e, 149\u0026ndash;172 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDaovisan, H. Responsible or Sustainable AI? Circular Economy Models in Smart Cities. \u003cem\u003eSustainability\u003c/em\u003e \u003cb\u003e18\u003c/b\u003e, 398 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBatty, M. \u003cem\u003eInventing Future Cities\u003c/em\u003e. (2018). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.7551/mitpress/11923.001.0001\u003c/span\u003e\u003cspan address=\"10.7551/mitpress/11923.001.0001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKwasi Anarfi, C., Shiel \u0026amp; Ross, A. Hill. The public\u0026rsquo;s perspectives of urban sustainability: A comparative analysis of two urban areas in Ghana. \u003cem\u003eHabitat Int.\u003c/em\u003e \u003cb\u003e111\u003c/b\u003e, 102496 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eReginald, G. \u0026amp; Golledge Geography and the disabled: A survey with special reference to vision impaired and blind populations. \u003cem\u003eTrans. Inst. Br. Geogr.\u003c/em\u003e \u003cb\u003e18\u003c/b\u003e, 63\u0026ndash;85 (1993).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDijk, J. \u003cem\u003eThe Digital Divide\u003c/em\u003e (Polity, 2020). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/asi.24355\u003c/span\u003e\u003cspan address=\"10.1002/asi.24355\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKitchin, R. The Real Time City༟Big data and smart urbanism. \u003cem\u003eGeoJournal\u003c/em\u003e 79, 1\u0026ndash;14 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRob Kitchin. Data-driven Urbanism. \u003cem\u003eData City\u003c/em\u003e. \u003cb\u003e13\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.4324/9781315407388\u003c/span\u003e\u003cspan address=\"10.4324/9781315407388\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVel\u0026aacute;zquez, R. et al. An Outdoor Navigation System for Blind Pedestrians Using GPS and Tactile-Foot Feedback. \u003cem\u003eAppl. Sci.\u003c/em\u003e \u003cb\u003e8\u003c/b\u003e, 578 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDavid Stea. \u003cem\u003eImage and Environment Cognitive Mapping and Spatial Behavior\u003c/em\u003e (Routledge, 2017). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.4324/9780203789155\u003c/span\u003e\u003cspan address=\"10.4324/9780203789155\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAjzen, I. The theory of planned behavior. \u003cem\u003eOrgan. Behav. Hum Decis. Process.\u003c/em\u003e \u003cb\u003e50\u003c/b\u003e, 179\u0026ndash;211 (1991).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEl-taher, F. E., Taha, A., Courtney, J. \u0026amp; Mckeever, S. A Systematic Review of Urban Navigation Systems for Visually Impaired People. \u003cem\u003eSensors\u003c/em\u003e \u003cb\u003e21\u003c/b\u003e, 3103 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eImrie, R. \u0026amp; Luck, R. Designing inclusive environments: rehabilitating the body and the relevance of universal design. \u003cem\u003eDisabil. Rehabil\u003c/em\u003e. \u003cb\u003e36\u003c/b\u003e, 1315\u0026ndash;1319 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eL\u0026auml;ttman, K., Olsson, L., Friman, M. \u0026amp; Fujii, S. Perceived Accessibility, Satisfaction with Daily Travel, and Life Satisfaction among the Elderly. \u003cem\u003eIJERPH\u003c/em\u003e \u003cb\u003e16\u003c/b\u003e, 4498 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eImrie, R. \u0026amp; Hall, P. \u003cem\u003eInclusiveDesign Designinganddevelopingaccessibleenvironments\u003c/em\u003e (Spon, 2001). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.4324/9780203362501\u003c/span\u003e\u003cspan address=\"10.4324/9780203362501\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMarion, A., Hersh \u0026amp; Johnson, M. A. \u003cem\u003eAssistive Technology for Visually Impaired and Blind People\u003c/em\u003eSpringer London,. (2008). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-1-84628-867-8\u003c/span\u003e\u003cspan address=\"10.1007/978-1-84628-867-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiaw, A. \u0026amp; Wiener, M. Classification and Regression by randomForest. \u003cem\u003eR News\u003c/em\u003e. \u003cb\u003e2\u003c/b\u003e, 18\u0026ndash;22 (2002).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCurl, A., Nelson, J. D. \u0026amp; Anable, J. Does Accessibility Planning address what matters? A review of current practice and practitioner perspectives. \u003cem\u003eRes. Transp. Bus. Manage.\u003c/em\u003e \u003cb\u003e2\u003c/b\u003e, 3\u0026ndash;11 (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNegm, H., De Vos, J. \u0026amp; Pot, F. El-Geneidy, A. Perceived accessibility: A literature review. \u003cem\u003eJ. Transp. Geogr.\u003c/em\u003e \u003cb\u003e125\u003c/b\u003e, 104212 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJames, J. \u0026amp; Gibson \u003cem\u003eThe Ecological Approach to Visual Perception\u003c/em\u003eNew York,. (2014). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.4324/9781315740218\u003c/span\u003e\u003cspan address=\"10.4324/9781315740218\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFadhlillah, F. Rethinking the Tactile Paving Installation System Based on the City Rhythm of Visually Impaired Pedestrians in Urban Networks. \u003cem\u003eJurnal Pembangunan Wilayah dan. Kota\u003c/em\u003e \u003cb\u003e20\u003c/b\u003e, (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlmoshaogeh, M. et al. A Review on Disability-Inclusive Public Transportation: Current Barriers and Prospects. \u003cem\u003eIEEE Access.\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e, 75769\u0026ndash;75786 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCaragliu, A. \u0026amp; Del Bo, C. F. Smart innovative cities: The impact of Smart City policies on urban innovation. \u003cem\u003eTechnol. Forecast. Soc. Chang.\u003c/em\u003e \u003cb\u003e142\u003c/b\u003e, 373\u0026ndash;383 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUnited Nations, General Assembly. \u003cem\u003eNew Urban Agenda\u003c/em\u003e. (2016). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://unhabitat.org/sites/default/files/2019/05/nua-english.pdf\u003c/span\u003e\u003cspan address=\"https://unhabitat.org/sites/default/files/2019/05/nua-english.pdf\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Hierarchical environmental semantic reconstruction, Visually impaired pedestrians, Stable and coherent mobility, Accessibility governance, Urban streets","lastPublishedDoi":"10.21203/rs.3.rs-9101465/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9101465/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eUrban street accessibility for visually impaired individuals has traditionally been evaluated through the lens of barrier-free infrastructure. However, this study demonstrates that mobility stability is determined not merely by the physical presence of facilities, but by the extent to which environmental cues are coherently perceived, interpreted, and translated into action.\u003c/p\u003e \u003cp\u003eUsing Shanghai as a case study, we analyzed 309 questionnaire responses to examine how environmental barriers and psychological factors jointly shape travel intention. A random forest model achieved 78% classification accuracy in predicting travel avoidance behavior, identifying psychological anxiety and tactile paving obstruction as the most significant predictors. The findings suggest that disruptions in the built environment influence mobility primarily through the psychological mediation of environmental uncertainty.\u003c/p\u003e \u003cp\u003eFurthermore, k-means clustering identified four distinct mobility profiles, designated as: Stable Traditional Users, Vulnerable Environment-Dependent Users, Technology-Assisted but Environmentally Sensitive Users, and Technology-Empowered Low-Anxiety Users. These groups exhibit significant variances in perceived environmental barriers, psychological anxiety, and assistive technology adoption.\u003c/p\u003e \u003cp\u003eThis study establishes a hierarchical framework linking environmental signals, perception, and behavior. The results reveal that mobility stability emerges through differentiated \"interruption-compensation\" pathways, advocating for a paradigm shift in accessibility governance from basic facility provision toward semantic reliability and digital-physical synchronization.\u003c/p\u003e","manuscriptTitle":"Hierarchical Environmental Semantic Reconstruction for Stable Mobility of Visually Impaired Pedestrians on Urban Streets in Shanghai","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-05 17:10:51","doi":"10.21203/rs.3.rs-9101465/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"7aaaf55f-9f64-4e5d-a024-cb5e2c97c910","owner":[],"postedDate":"April 5th, 2026","published":true,"recentEditorialEvents":[{"type":"decision","content":"Rejected","date":"2026-04-30T04:40:16+00:00","index":"","fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":65622101,"name":"Earth and environmental sciences/Environmental social sciences"},{"id":65622102,"name":"Biological sciences/Psychology"},{"id":65622103,"name":"Social science/Psychology"}],"tags":[],"updatedAt":"2026-04-30T04:54:39+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-05 17:10:51","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9101465","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9101465","identity":"rs-9101465","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}