Intelligent design empowers spatial aesthetics: construction of a man-machine collaborative innovation model for environmental art and case verification

Intelligent design empowers spatial aesthetics: construction of a man-machine collaborative innovation model for environmental art and case verification | 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 Intelligent design empowers spatial aesthetics: construction of a man-machine collaborative innovation model for environmental art and case verification Shaoqing Wang, Jian Yao This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7464844/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 Driven by intelligent technologies, environmental art design is undergoing a paradigm reconstruction from experience dependence to a data-knowledge dual-engine support model. Aiming at the structural defects of the traditional paradigm, such as efficiency bottlenecks, rigid creative boundaries, and sluggish dynamic responses, this study constructs an intelligent design model based on the logic of human-machine symbiosis. Through the integration of three technical modules: the spatial semantic analysis of the generative adversarial network, the multi-objective dynamic optimization of the parametric system, and the immersive collaboration of the augmented reality interface, a closed-loop innovation chain of aesthetic feature vectorization expression - algorithm generation - manual adjustment is formed. Empirical research shows that in typical scenarios such as the activation of cultural heritage and intelligent complexes, this model can increase the efficiency of creative plan generation by 42%, achieve an 82.5% consistency in the evaluation of interdisciplinary experts, enhance the intensity of user emotional resonance by 60% compared with traditional methods, and reduce the construction plan change rate by 55%. This technical framework not only reveals the computable interaction mechanism between algorithm logic and aesthetic laws, but also provides a transformation path with both cognitive breakthroughs and engineering implementation potential for the environmental art field by establishing dynamic weight allocation and ethical review rules. Its methodological system can be extended to the sustainable construction practices of complex systems such as urban renewal and digital cultural heritage protection. Physical sciences/Engineering Physical sciences/Mathematics and computing Intelligent design Spatial aesthetics Human-machine Collaboration Generative adversarial network Multi-objective optimization Multimodal interaction Aesthetic computing Figures Figure 1 Figure 2 Figure 3 Introduction Research background In recent years, the deep penetration of artificial intelligence and big data technologies is triggering a paradigm reconstruction in the field of environmental art design 1 . Intelligent algorithms represented by generative design and deep learning are gradually breaking through the capacity boundaries of traditional tools in spatial form deduction and aesthetic law analysis. Through the feature deconstruction and non-linear correlation mining of a vast number of cases, they have achieved a leap from static plan output to dynamic creative generation. Especially in the scenario of functional-aesthetic coupling optimization in complex spaces, algorithm-driven high-dimensional parametric modeling can simultaneously handle lighting relationships, material properties, and ergonomic constraints, significantly improving the accuracy and systematicness of design iteration and providing quantifiable technical support for multi-objective decision-making. However, the field of environmental art design still faces the dual challenges of experience dependence and technical adaptability. In current industry practices, designers’ shaping of spatial aesthetics is mostly based on individual aesthetic qualities and the accumulation of project experience. The creative mode dominated by subjective judgment not only leads to a long scheme generation cycle and low resource allocation efficiency, but also makes it difficult to meet the high-level requirements for dynamic adaptability in emerging scenarios such as urban renewal and smart buildings. At the same time, the professional barriers among architecture, landscape, and interior design in the engineering implementation stage make it difficult to effectively integrate interdisciplinary data, often resulting in the separation of aesthetic expression and functional logic, and exacerbating the construction rework rate and energy consumption. Against this backdrop, there is an urgent need to construct a collaborative framework of engineering rationality and artistic sensibility for the innovative path of spatial aesthetics. On the one hand, as a material carrier, space needs to meet the requirements of physical properties such as structural stability and environmental compatibility. On the other hand, as a cultural medium, it must also fulfill the social functions of carrying emotional resonance and transmitting aesthetic values. The one-way thinking of “technology first” or “form supremacy” in traditional design methods can no longer meet the demands of the era that emphasizes both people-oriented orientation and sustainable development. By introducing a human-machine collaborative mechanism, the intelligent system can expand the creative boundaries relying on its data mining capabilities, while designers can make aesthetic adjustments and value judgments on the algorithm output with critical thinking 2 . The in-depth interaction between the two will drive the design process to shift from linear progression to closed-loop optimization, ultimately achieving an organic unity of technological empowerment and humanistic care. Research significance The integrated innovation of intelligent technology and environmental art design is essentially a dual expansion of the ontology and methodology of design. From a theoretical dimension, transforming the formal rules and emotional experiences of spatial aesthetics into a computable and optimizable parameter system 3 not only breaks through the cognitive paradigm dominated by qualitative descriptions in traditional design research, but also provides a systematic framework for interdisciplinary knowledge integration by establishing a mapping relationship of “aesthetic features - algorithm logic - human factor feedback”. This aesthetic computing model based on data density and algorithm complexity can effectively decode the implicit laws in the evolution of historical styles and capture the dynamic changes in the aesthetic preferences of contemporary users. Thus, it builds a two-way verification channel between quantitative analysis and qualitative judgment, promoting the paradigm innovation of design theory from empirical induction to deductive reasoning. In the practical dimension, the collaborative design mechanism empowered by algorithms is reconstructing the value chain of environmental art creation. By embedding multi-objective optimization algorithms and real-time rendering technologies, the intelligent system can quickly generate candidate solutions that balance functional rationality and visual expressiveness 4 . It also incorporates dynamic variables such as user behavior data and environmental sensor information into the iterative cycle, significantly enhancing the adaptability of design solutions to complex scenarios. This data-driven decision-making model not only reduces the time and resource consumption brought about by the traditional trial-and-error method but also breaks the information silos among sub-fields such as architecture, landscape, and interior design by establishing a cross-professional collaboration platform. As a result, the creation of spatial aesthetics has shifted from discrete fragmented operations to integrated innovation of full-life-cycle management. Especially in emerging fields such as smart city construction and digital protection of cultural heritage, the human-machine collaborative model provides a scalable solution for balancing technological innovation and cultural inheritance 5 . Its dynamic response characteristics are more in line with the requirements of the times for intensive utilization of resources under the goal of sustainable development. Research questions The construction of the collaborative paradigm between intelligent systems and human designers essentially aims to address the dialectical relationship between machine logic and humanistic values in environmental art creation. The primary challenge lies in how to design interaction rules and responsibility boundaries to make the efficient computing ability of algorithms complement rather than compete with the aesthetic judgment of designers. This requires the establishment of a phased and hierarchical collaborative mechanism: in the concept generation stage, the intelligent system needs to break through the stylistic limitations of traditional parametric tools and capture unstructured aesthetic features through generative adversarial networks 6 ; while in the scheme refinement stage, designers need to take the lead in interpreting cultural symbols and constructing spatial narratives to ensure that the candidate schemes output by the algorithm have adaptable creative extensibility. The collaborative efficiency of the two not only depends on the friendliness of the technical interface but also relies on the accuracy of the algorithmic translation of the tacit knowledge of design thinking. Furthermore, the evaluation of the improvement effect of spatial aesthetic quality needs to break through the ambiguity of traditional subjective evaluation and construct a composite index system that integrates objective data and humanistic criteria. This involves the quantitative connection of three dimensions: First, through the style classification model and emotion computing technology 7 , the aesthetic features are deconstructed into measurable parameters such as color contrast and morphological complexity entropy. Second, an expert experience database is established, and the analytic hierarchy process is used to determine the weight distribution of each aesthetic element. Third, eye-tracking and physiological signal monitoring are introduced to capture users’ subconscious feedback on the spatial atmosphere 8 . Only by achieving the multi-source alignment of algorithm predicted values, expert scores, and user perception data can an intelligent design evaluation benchmark with industry consensus be formed. In addition, when the human-machine collaborative model is applied to complex scenarios such as the renewal of urban complexes and historical blocks, its robustness needs to be ensured through a multi-scale verification framework. At the technical level, it is necessary to test the generalization ability of the algorithm in data-sparse scenarios. For example, the style characteristics of existing projects can be adapted to heterogeneous cultural contexts through transfer learning. At the application level, it is necessary to evaluate the response sensitivity of the model to dynamic variables, including real-world constraints such as sudden changes in functional requirements and aesthetic conflicts among multiple stakeholders. Only by ensuring that the algorithm strikes a balance between stability and adaptability can the potential of the intelligent design system to transform from an experimental tool to an engineered platform be confirmed. Research objectives To achieve the intelligent transformation of the environmental art design paradigm, it is necessary to establish the research objective orientation from the dual perspectives of technological innovation and value reconstruction. The primary task is to construct the technical ontology of the man-machine collaborative innovation model. By deconstructing the cognitive decision-making chain in the design process, the role division of the intelligent system and designers at different stages can be clarified. At the technical path level, it is necessary to integrate the style transfer ability of the generative adversarial network, the constraint solving efficiency of the multi-objective optimization algorithm, and the immersive tuning function of virtual reality interaction to form a closed-loop architecture of “data perception - plan generation - dynamic optimization”. The design of interaction rules focuses on the dynamic allocation mechanism of creative control rights. For example, the algorithm is given higher autonomy during the spatial form topological optimization stage, while the priority of manual intervention is strengthened in the cultural semantics injection link, so as to achieve the gradient integration of machine logic and human creativity. On this basis, the research objective further focuses on the verification of the practical effectiveness of the model. A scientific evaluation system needs to be established through case-based empirical studies in multiple scenarios and at multiple scales. Regarding the aesthetic expression dimension, two types of projects with vastly different styles, namely the revitalization of historical blocks and futuristic pavilions, are selected. The differences in the algorithm’s expressiveness in cultural symbol extraction and avant-garde form interpretation are comparatively analyzed. At the functional adaptation level, the real-time feedback of sensor data on the spatial utilization efficiency of commercial complexes is used to verify the response accuracy of the intelligent system to dynamic variables such as passenger flow density and energy consumption indicators. For the verification of user experience, eye movement trajectory heatmaps and emotional semantic analysis are relied upon to quantitatively evaluate the improvement amplitude of the human-machine collaboration scheme in terms of spatial narrative coherence and emotional resonance intensity. This full-life-cycle empirical framework will reveal the transformation boundaries and adaptation conditions of intelligent design models from theoretical construction to engineering implementation. Literature review Research progress of intelligent design The evolution of intelligent design technology is profoundly reconstructing the methodological system of architectural form generation and aesthetic analysis 9 . Generative design, by embedding genetic algorithms and topological optimization principles, transforms the traditional form creation relying on empirical intuition into a multi-objective-driven parametric solution process. Its core advantage lies in simultaneously handling the complex coupling relationships among structural performance, material efficiency, and spatial experience. Taking large public buildings as an example, through iterative calculations, the algorithm can achieve a dynamic balance between load-bearing efficiency and curved surface aesthetics, generating irregular structure schemes that are both mechanically reasonable and visually appealing. This design paradigm of automatically optimizing based on constraints significantly improves the solution efficiency of complex engineering problems and the diversity of design schemes. Meanwhile, the intervention of deep learning technology has further expanded the aesthetic analysis dimension of intelligent design. The theoretical foundation of this technological evolution can be traced back to the paradigm breakthrough of aesthetic computing - it quantifies formal rules and emotional experiences into high-dimensional computable parameters, constructs an interactive interface between artistic creation and algorithmic logic, and promotes the transformation of environmental art design from an experience-driven paradigm to a data-knowledge dual-engine supported one. By extracting the style features of the architectural historical image library through convolutional neural networks 10 , the system can establish quantitative representation models of aesthetic laws such as the vertical rhythm of Gothic style and the simple proportion of modernism, and achieve the contemporary translation of traditional vocabulary with the help of style transfer algorithms. This data-driven feature reconstruction ability enables intelligent systems not only to imitate established styles, but also to generate novel forms that break through the boundaries of experience through exploration in the latent space. For example, regional cultural symbols can be deconstructed into topologically variable basic units and then recombined into innovative forms that meet contemporary functional requirements. However, the existing technology still faces a structural mismatch between the generated results and humanistic aesthetic standards. Especially in the deep aesthetic dimensions such as cultural metaphor conveyance and emotional resonance intensity, algorithmic logic is still difficult to completely replace the critical thinking of human designers. Technological challenges of spatial aesthetics The penetration of intelligent design technology into the field of spatial aesthetics has always faced the structural tension between machine logic and humanistic values. Although the algorithm-generated solutions perform excellently in terms of morphological complexity and engineering rationality, there remains an essential contradiction in the degree of fit between their aesthetic output and the common human aesthetic consensus. This contradiction stems from the context-dependence of aesthetic judgment: The style association models established by algorithms through massive data training often simplify cultural symbols into probability distributions of visual features, but it is difficult to capture unstructured elements such as the sense of ritual and collective memory in the local context. For example, in the scenario of historical block renewal, the generated solutions may accurately reproduce the proportional scale of traditional architecture, but lose the time narrative carried by the aging traces of materials, resulting in the spatial experience falling into “precise paleness”. More importantly, the current aesthetic evaluation datasets mostly originate from the Western modernist paradigm. This cultural bias may cause the algorithm to overemphasize geometric order when generating irregular forms, suppressing the expression of non-linear characteristics such as “blank space” and “artistic conception” in Eastern aesthetics. When dealing with dynamic spatial requirements, the rigid characteristics of the technical framework of parametric tools are becoming increasingly prominent 11 . Although topological optimization algorithms can efficiently handle the generation of forms under static constraints, when faced with dynamic variables such as sudden changes in users’ behavior patterns and real-time changes in environmental lighting, the existing toolchains often exhibit system inertia. This limitation is essentially a contradiction between the “priori setting” methodology of parametric modeling and the “emergent characteristics” of the real world: designers need to pre-define all the parameters that may affect the spatial form and their weight of influence, while the dynamic interactions in reality often produce associative effects beyond the pre-set scope. Taking urban public spaces as an example, real-time data such as pedestrian flow density and types of social activities should drive the adaptive adjustment of the spatial interface. However, due to the lack of an online learning mechanism in traditional parametric systems, it is difficult to convert sensor data streams into form adjustment instructions, resulting in dynamic design remaining at the conceptual stage. This tool limitation not only restricts the scene response ability of spatial aesthetics but also hinders the evolution of human-computer collaborative design towards full-life-cycle management. Paradigm of human-machine collaborative design The evolution of the human-machine collaborative design paradigm is redefining the boundaries and implementation paths of creativity 12 . Within the theoretical framework of role division of labor, the collaboration between designers and intelligent systems is not simply a task segmentation, but a differential empowerment based on cognitive advantages: designers take the lead in value judgment and cultural decoding, and are responsible for setting aesthetic goals and ethical constraints; intelligent systems, on the other hand, give full play to their computational advantages in hyperparameter space exploration and multi-objective optimization, and transform abstract aesthetic demands into executable morphological strategies. This collaborative model of “humans setting the direction - machines expanding possibilities” is particularly evident in the digital restoration of historical buildings. Designers endow the cultural context of space restoration through semantic networks, and algorithms generate technical solutions that conform to structural specifications and style continuity based on this. The two continuously calibrate the creative trajectory through a real-time feedback loop to ensure that technical rationality always serves the expression of humanistic values. The introduction of interactive evolutionary algorithms provides a dynamic creative tuning mechanism for human-computer collaboration 13 . This technology converts the designer’s subjective evaluation into a fitness function, enabling the algorithm to continuously capture human preference characteristics during the iterative process, and thus achieve the progressive optimization of the generation direction. Taking commercial space design as an example, the designer’s aesthetic preference scores for the initial set of design schemes are encoded as the retention probabilities of morphological genes, driving the algorithm to strengthen key features such as curve softening and material contrast in subsequent iterations, while automatically filtering out structural proposals that violate ergonomics. This semi-automated evolutionary process not only retains the human’s ultimate right to decide on aesthetic quality but also breaks through the mindset of traditional design processes through the machine’s parallel computing capabilities. More importantly, the interactive evolutionary algorithm constructs an interpretable creative evolution path, enabling designers to trace the generation logic of specific aesthetic characteristics, thereby establishing a verifiable cognitive bridge between perceptual judgment and algorithmic decision-making. Research gaps While promoting the innovation of technical tools in current intelligent design research, there are still crucial discontinuities in its methodological system. The primary limitation lies in the disconnection between the aesthetic evaluation dimension and the algorithm generation process: most models simplify aesthetics to the optimization of the probability distribution of morphological parameters, but fail to establish a quantitative index system that is deeply associated with regional culture and emotional experience. This technological tendency results in a paradox where the generated solutions often exhibit “formal correctness” but “meaning vacuity”. For example, although algorithms can accurately fit the proportional characteristics of new Chinese-style architecture, they are unable to decode the metaphorical relationship between the upturned angle of the eaves and the spirit of the place, and it is even more difficult to quantitatively evaluate the intensity of cultural identity that users experience when they are in the space. The lack of such an evaluation system turns the intelligent design system into a style collage tool rather than a true aesthetic innovation engine. A more profound limitation lies in the one-sidedness of the case verification paradigm. Most existing studies focus on proving the superiority of algorithms in technical indicators such as generation speed and structural rationality, but rarely construct an interdisciplinary evaluation framework that encompasses art value criticism and social acceptance analysis. Taking parametric public art installations as an example, research literature generally emphasizes the topological optimization efficiency of form generation algorithms, but ignores core art value dimensions such as the dialogue ability between the installation and the urban context and public participation. This verification bias may mislead the direction of technology research and development, causing intelligent design to fall into the dilemma of being “technically feasible” but “culturally silent”. Even more severe is that when algorithm-generated solutions involve sensitive scenarios such as the renovation of historical blocks, the lack of a social acceptance evaluation mechanism will amplify the technical ethics risks, leading to a cultural identity crisis for the design solutions during the implementation phase. Materials and methods Data source and processing The acquisition and processing mechanism of research data is a fundamental task to support the construction of the human-machine collaborative model. The construction of the spatial aesthetics feature database follows the principles of scene taxonomy, covering 10 types of typical environments such as commercial complexes, cultural exhibition halls, and historical blocks. Through 3D laser scanning and high-precision texture mapping technologies, more than 5,000 design schemes are transformed into computable data entities. Each case is subject to multi-layer semantic annotation: in the style dimension, a convolutional neural network is used to automatically identify the regional feature index 14 . The material properties are double-verified through spectral reflectance and tactile simulation parameters, and the lighting data at different times is reconstructed for the light and shadow relationship relying on the radiosity algorithm. To ensure the consistency of the annotation logic, an expert-algorithm collaborative verification mechanism is adopted. Scholars in the fields of architectural history and environmental psychology calibrate the feature labels in the cultural context to eliminate the meaning distortion caused by simple visual feature matching. The quantitative collection of users’ aesthetic preferences is achieved relying on the multi-modal perception technology system 15 . The eye-tracking system records the fixation point heat map of observers during virtual space roaming at a sampling rate of 250Hz, and analyzes the distribution of visual attractiveness through the dwell time of gaze and the complexity of saccade paths. At the same time, a wrist-worn physiological sensor synchronously collects galvanic skin responses and heart rate variability to establish an association model between spatial atmosphere stimuli and emotional arousal levels. To improve the ecological validity of the data, the experimental scenario uses augmented reality technology to overlay real environmental lighting and acoustic parameters, ensuring that the feedback from participants reflects the real spatial experience. After time series alignment and outlier removal of the original data, it is transformed into feature vectors with spatial grid encoding, and finally a multi-dimensional evaluation matrix that couples style preferences, functional requirements, and emotional feedback is formed, providing a training benchmark for the adaptive optimization of the intelligent system. Technical tools The construction of the technical toolchain follows the collaborative logic of “generation - optimization - interaction”, aiming to open up the value transfer channel from algorithmic computing to manual decision - making. The generation module adopts the improved StyleGAN3 architecture 16 . Through adaptive artifact suppression and latent space disentanglement technologies, it significantly improves the output quality of irregular building forms. During the training phase, this module injects knowledge in the field of architecture, deconstructs the historical case library into style gene segments, enabling the generator to recombine innovative solutions that not only conform to formal aesthetics but also avoid copyright disputes. Compared with traditional generative adversarial networks, the improved architecture increases the diversity of solutions by 37% while maintaining style continuity, effectively alleviating the constraints of mode collapse on creative exploration. On this basis, the Multi-Objective Genetic Algorithm (MOGA) constructs a dynamic balance mechanism between functionality and aesthetics 17 . The algorithm encodes indicators such as structural rationality, construction cost, and visual attractiveness into a multi-dimensional fitness function. Through the elitist retention strategy and non-dominated sorting, it screens out the optimal solution set on the Pareto front. In particular, for the conflicting constraints in spatial design, the algorithm introduces a dynamic weight adjustment mechanism. When the user feedback data shows a significant deviation during the scheme refinement stage, it automatically reduces the priority of the morphological complexity index to ensure that the optimization direction always targets the core requirements. This flexible optimization strategy increases the efficiency of scheme iteration by 58% compared to traditional methods, while maintaining the baseline standard of aesthetic quality. Furthermore, implement embodied operations of human-machine collaboration through a virtual reality interactive platform 18 . This platform integrates a ray tracing rendering engine and a physical simulation system, which can real-time present the material reflection characteristics and spatial scale perception of design schemes under real lighting conditions. Designers directly grasp and stretch the key parameter nodes of virtual models through the gesture control interface, and the system simultaneously updates morphological indicators and performance data such as acoustics and thermal engineering. This what-you-see-is-what-you-get interactive mode not only reduces the time-consuming of scheme adjustment to 1/5 of that of traditional CAD tools, but also constructs an interpretable creative evolution path through the parameter operation history tracing function. The platform is specifically equipped with a collaborative mode, allowing the algorithm to continuously provide alternative scheme branches during the designer’s modification process, thus forming a two-way inspiring creative dialogue. Model construction The technical architecture of the human-machine collaborative innovation model follows the construction logic of “data-driven - intelligent emergence - value closed-loop”, and realizes the digital translation of design thinking through a hierarchical and progressive structure. At the data layer, the multi-modal feature extraction engine deconstructs spatial aesthetic elements into style vectors, material fingerprints, and light and shadow matrices. At the same time, it constructs a knowledge graph in a cross-cultural context to form a quantifiable and comparable design semantic network 19 . The algorithm layer adopts a cross-modal attention mechanism to achieve the dynamic alignment of text requirement descriptions, 3D point cloud data, and user physiological feedback signals 20 . Through the gated recurrent unit, it controls the feature weight distribution in different design stages to ensure that functional constraints and aesthetic goals achieve the Pareto frontier optimal solution in the parameter space. The application layer deploys a lightweight inference engine to convert the high-dimensional algorithm output into interactive BIM model components in real time, and establishes a two-way feedback channel to capture the designer’s adjustment intentions, forming a continuously evolving design knowledge base. The core of the collaborative mechanism lies in the formulation of a dynamic rights and responsibilities allocation strategy 21 . At the initial stage of design, the system generates a set of candidate solutions that go beyond the constraints of conventional experience through potential space exploration. It uses style interpolation technology to present a continuous spectrum from conservative to radical, inspiring designers to break through mental stereotypes. When the design plan enters the deepening stage, the collaborative rules automatically switch to the “human-dominated, machine-assisted” mode: the algorithm turns to conducting sensitivity analysis of detailed parameters, predicting the impact gradient of material replacement or morphological fine-tuning on the overall aesthetic evaluation, while designers focus on the topological adaptation of cultural symbols and the construction of spatial narratives 22 . In the final decision-making process, the built-in ethical review module of the system will detect whether the design plan has the risk of cultural stereotypes or characteristics of group exclusion. At the same time, it provides an interpretability report to clarify the algorithmic basis for key design decisions, ensuring the coexistence and evolution of technical rationality and humanistic values in spatial construction. Case verification design The case verification design follows the principle of “scenario differentiation - indicator multi - dimensionalization - process traceability”. The universality of the model’s effectiveness is verified by ensuring full coverage of typical spatial types. The experimental objects focus on three types of scenarios with representative challenges: Urban public spaces mainly verify the model’s response ability to social inclusiveness and functional complexity; commercial complexes test the balance accuracy of human - machine collaboration in terms of economic benefits and user experience 23 ; cultural exhibition halls mainly examine the possible boundaries of the integration of traditional symbol translation and avant - garde technologies. To ensure the ecological validity of the experiment, each type of scenario includes two project types: new construction and renovation. And the geographical distribution of cases is limited to cover different climate zones and cultural circles, so as to test the adaptability of the model under cross - environmental constraints. The construction of the evaluation index system breaks through the single dimension of the traditional technology-oriented approach 24 . Innovatively, it incorporates efficiency gain, professional consensus, and user perception into a unified analysis framework. The design cycle shortening rate calculates the total time compression ratio and the efficiency improvement value of key decision points by comparing the time-consuming differences between the human-machine collaborative model and the traditional process in key stages such as concept generation and scheme deepening 25 . For the consistency of expert scores, an interdisciplinary review group is formed. The double-blind test method is used to independently score indicators such as the cultural fit and technical rationality of the scheme. The Cohen’s Kappa coefficient is used to quantify the judgment consensus among architectural historians, structural engineers, and art curators. A threshold of ≥ 0.75 is set as the baseline for the reliability of aesthetic evaluation. The user satisfaction survey abandons the single overall score. The Likert 5-point scale is used to collect data on sub-dimensions such as spatial readability, emotional resonance, and functional convenience in a hierarchical manner. The principal component analysis is used to extract the latent variables affecting satisfaction and establish a mapping relationship with the algorithm predicted values. All evaluation processes record operation logs and decision-making paths to ensure that the verification results are reproducible and auditable. To quantify the intervention effect of the intelligent design model, Cohen’s d value is used for effect size analysis. Its calculation formula is: The calculation method of the combined standard deviation SDmerge is as follows: In the formula: M experimental group represents the mean of the data of the experimental group; M control group represents the mean of the data of the control group; SD1 represents the standard deviation of the data of the experimental group; SD2 represents the standard deviation of the data of the control group; n1 represents the sample size of the experimental group; n2 represents the sample size of the control group. The interpretation criteria for the effect size refer to Cohen’s guidelines: small (d = 0.2), medium (d = 0.5), large (d = 0.8). Results Model performance The comparative test data based on three typical scenarios show that the human-machine collaborative model has significantly improved in terms of design efficiency and aesthetic quality. In the efficiency dimension, by recording the time cost and the frequency of plan iteration in the conceptual design stage, it is found that compared with the traditional workflow, the intelligent system increases the generation efficiency of creative plans by 42% (the average value drops from 18.3 hours per plan to 10.6 hours), and reduces the number of design iterations by 60% (the average value is compressed from 7.2 times to 2.9 times). This gain comes from the parallel computing ability of the generation module and the constraint pre-screening mechanism of the optimization algorithm, effectively avoiding the invalid path dependence in manual plan exploration (as shown in Figure 1). At the aesthetic evaluation level, a composite evaluation method that combines double-blind expert review and user perception is adopted. The aesthetic consistency score of the model output scheme reaches 82.5%, which is a 51.9% increase compared to 54.3% of the traditional method. This leap benefits from the dynamic alignment ability of the cross-modal attention mechanism to cultural context and functional requirements (as shown in Table 1). Table 1. Comparison of aesthetic evaluation consistency (%) Evaluation dimension Traditional solution Human-machine collaboration solution Δ value Cohen’s d p value Emotional resonance degree 3.7±0.9 4.5±0.6 +0.8 1.05 <0.001*** Functional convenience 3.4±1.1 4.2±0.7 +0.8 0.87 0.002** Visual appeal 3.8±0.8 4.4±0.5 +0.6 0.90 0.013* Spatial readability 3.5±1.0 4.1±0.6 +0.6 0.73 0.021* Case empirical research The application verification of the human-machine collaborative model in typical scenarios reveals the differentiated value paths of intelligent technologies enabling spatial aesthetics. In the case of cultural exhibition hall design, the generation module outputs 8 stylized structural schemes such as Neo-Futurism and Deconstructivism through style interpolation technology. Among them, the theme of “broken line metaphor”was selected by the designer because it fits the industrial heritage context of the site. The system then initiates multi-physical field coupling optimization: based on the flow line analysis of pedestrian flow simulation data, the width of the main passage is dynamically adjusted from the initial plan of 3.2 meters to an elastic range of 2.8 - 4.5 meters, enhancing the fun of spatial exploration while ensuring evacuation efficiency; for the lighting parameter optimization, a non-uniform B-spline curve is introduced to control the light attenuation gradient, keeping the illuminance of the exhibition wall within the optimal viewing range of 150 - 200 lux, and the power density is reduced by 22% compared to traditional designs. This dynamic tuning mechanism liberates designers from technical details and allows them to focus on the construction of spatial narratives (as shown in Figure 2). Compared with the morphological innovation challenges of cultural exhibition halls, the case of the atrium renovation of commercial spaces highlights the value transformation ability driven by user needs. By analyzing 2,350 sets of user eye movement data and heart rate variability records, the algorithm identified a preference cluster for “natural light and shadow”, and based on this, an optimized plan with a parametric wooden grille dome was generated. The size of the grille units changes dynamically according to the solar trajectory (the porosity is 15% at noon in summer → 45% in the morning in winter). While controlling the glare index, it allows the spatial illuminance to fluctuate naturally within the comfortable range of 300 - 500 lux. During the construction phase, the robot prefabrication and assembly technology was adopted to control the error of 7,200 special-shaped components within ±1.5 mm. Finally, the project cost was reduced by 18% (as shown in Table 2), and the average user stay time increased by 24 minutes. Table 2. Comparison of the benefits of commercial space renovation indicator Traditional solution Human-machine collaboration solution Rate of change Construction cost (in ten thousand yuan) 850 697 -18.0% Average daily passenger flow (person-times) 2300 2890 +25.7% User satisfaction (on a 5-point scale) 3.6 4.3 +19.4% Energy consumption intensity（kWh/m²） 115 93 -19.1% Comparison between subjective and objective evaluations The validation of the effectiveness of the human-machine collaborative model requires simultaneously examining the objective accuracy of algorithm predictions and the matching degree of users’ subjective perception. The correlation analysis between expert scores and algorithm predicted values shows that there is a significant positive correlation between the two in the evaluation dimension of spatial aesthetic quality (Pearson’s r = 0.79, p < 0.01), indicating that the intelligent system can effectively capture the core aesthetic elements in professional evaluations. The distribution trend of the scatter plot reveals that when the algorithm predicted value is higher than 80 points (as shown in Figure 3), the standard deviation of expert scores narrows to ±3.2 points, indicating that the model has higher judgment stability in the identification of high-quality solutions. The results of the user satisfaction survey further confirm the practical value of the model. The human-machine collaboration solution is significantly superior to the traditional design method in sub-dimensions such as emotional resonance (4.5 vs. 3.7) and functional convenience (4.2 vs. 3.4) (independent samples t-test, p<0.001). Moreover, the standard deviation of overall satisfaction decreased from 0.82 in the control group to 0.51, reflecting that intelligent technology effectively reduces the evaluation divergence caused by individual aesthetic differences (as shown in Table 3). Table 3. Comparison of user satisfaction by dimension (on a 5-point scale) Evaluation dimension Traditional solution Human-machine collaboration solution Δ value p value Emotional resonance degree 3.7±0.9 4.5±0.6 +0.8 <0.001*** Functional convenience 3.4±1.1 4.2±0.7 +0.8 0.002** Visual appeal 3.8±0.8 4.4±0.5 +0.6 0.013* Spatial readability 3.5±1.0 4.1±0.6 +0.6 0.021* Discussion Model effectiveness The technical effectiveness of the human-machine collaboration model is reflected in its ability to bidirectionally expand the design cognitive framework. By deconstructing the implicit knowledge graph in historical cases, the intelligent system can establish a topological association network of cross-temporal and cross-cultural symbols. For example, the mechanical structure of the dougong in traditional architecture is deconstructed into a node connection rule base and then reorganized into an overhanging form that conforms to modern mechanical principles. This generation mechanism of non-explicit associations breaks through the time scale of designers’ individual experience and the spatial and cultural boundaries. In the digital reconstruction case of the New Museum of Suzhou, the algorithm-generated variant schemes of the eaves tile patterns through potential space exploration not only maintain the morphological rhythm of the Coiled chi-dragon patterns of the Warring States Period from the Warring States period but also adapt to the assembly logic of contemporary glass curtain walls through topological optimization. Such schemes were initially judged by designers as unconventional options because they exceeded the scope of traditional experience. However, their cultural translation effectiveness was confirmed after parametric verification. Meanwhile, the model ensures the decision-making dominance of humanistic values through a dynamic permission allocation mechanism. During the scheme deepening stage, the system automatically marks the culturally sensitive elements in the schemes generated by the algorithm (such as the spatial proportion of religious symbols, the deviation degree of historical authenticity, etc.), triggering the manual review process. Taking the renewal design of the Xi’an Small Wild Goose Pagoda area as an example, although the digital simplification scheme of the Tang-style roof ridge proposed by the algorithm passed the structural rationality verification, it was vetoed by the designer based on the data of historical literature research and the parameter rollback was initiated because the upturned angle of the eaves deviated from the threshold of the characteristics of the Tang Dynasty architectural style. This collaborative paradigm of “machine inspiration - human judgment” not only avoids the risk of style convergence caused by algorithmic hegemony, but also provides empirical support beyond the individual knowledge reserve for the designer’s critical decision-making through the cross-cultural knowledge retrieval function of the intelligent system, ultimately achieving the dialectical unity of technological innovation and cultural inheritance. Comparison with traditional methods The core breakthrough of the human-machine collaborative model compared with traditional environmental art design methods lies in its dynamic response ability to complex constraint systems and the depth of demand mining. In the traditional process, designers often handle hard constraints such as energy conservation and accessibility through linear superposition, resulting in multi-objective optimization falling into the local optimum trap. For example, in the aging-friendly renovation of old communities, when manually balancing the lighting requirements and thermal performance, conventional parametric tools can only generate a limited number of approximate solutions. In contrast, this model introduces implicit constraint modeling technology, encoding regulatory provisions and regional climate data into an adaptive weight matrix, enabling the algorithm to simultaneously optimize 23 parameters such as the angle of sunshading components and the turning radius of wheelchairs during the plan generation stage, and achieving the automatic approximation of the Pareto front. A more significant advantage lies in the ability to capture long-tail demands: Through the sentiment analysis of user behavior data and social media comments, the model can identify the needs of special groups that account for less than 5% (such as the spatial safety preferences of autistic children), and generate customized design elements such as high window sills and soft partitions accordingly. This is difficult to achieve in the traditional expert experience-driven model. At the same time, through the dynamic weight allocation mechanism, this model can real-time adjust the optimization priorities of function, aesthetics, and cost during the scheme deepening stage (such as automatically increasing the weight of modular assembly during the construction stage), enabling the identification and resolution of design conflicts to be advanced to the virtual verification link. As a result, the change rate of the construction plan is reduced from the industry average of 32–14.5%. This data confirms the role of dynamic weight allocation in integrating the design-construction value chain - the algorithm not only generates solutions but also anticipates implementation risks through the flexible adaptation of constraints, significantly reducing the marginal cost of project iteration. However, the model still faces structural challenges in terms of generation quality when dealing with small-sample scenarios. When the scarcity of training data and the complexity of design tasks create dual pressures (such as ultra-low-frequency projects like the renovation of Qiang watchtowers or burn rehabilitation centers), the algorithm may produce abnormal solutions with cultural context misalignment or functional logic conflicts. For example, in the case of the renovation of a southwestern ethnic minority village, due to the lack of samples of stilt building structures in the system, the generated elevated floor structure met modern seismic resistance specifications, but misapplied Dai architectural symbols to a Miao settlement. This reveals the shallow association defect of existing technologies with domain knowledge. There is an urgent need to construct a knowledge graph system that integrates unstructured knowledge such as construction methods and craftsmen’s formulas. In the future, through domain knowledge distillation technology, expert experience can be transformed into a tunable rule loss function, enabling the algorithm to maintain cultural adaptability even with a small number of samples, thus truly achieving the deep integration of intelligent design and humanistic values. Reflection on technological ethics The ethical risks of intelligent design technology coexist with its innovative potential. The problem of algorithmic bias manifests as a crisis of implicit style convergence in the field of spatial aesthetics. When the training dataset overly relies on cases of International Style modernist architecture, the generative system may generalize the visual principle of “less is more” into a universal aesthetic standard, resulting in regional cultural symbols being reduced to superficial decorative elements. For example, a clustering analysis of 48 intelligent generation plans for traditional village renewal found that 76% of the plans adopted a combination of standardized pitched roofs and glass curtain walls, while the utilization rate of local materials was less than 12%. This implicit bias essentially constitutes a technological and cultural hegemony in the digital age. Therefore, it is necessary to embed a diversity guarantee mechanism in the algorithm architecture, including establishing a cultural sensitivity detection module - identifying the density distribution of style features in the plan through a semantic network. When the probability of a certain style exceeds the preset threshold, the potential space perturbation algorithm is automatically triggered to force the exploration of the non-dominated solution region, thereby maintaining the diversity balance of the creative ecosystem. The deeper impact of technology ethics is reflected in the role reconstruction of the design subject. When the intelligent system undertakes more than 60% of the tasks of form generation and performance optimization, the core ability of designers is shifting from spatial form shaping to value strategy formulation. In the design of a certain science and technology innovation center in Xiongan New Area, the team is no longer confined to single form deliberation. Instead, by adjusting the algorithm weight matrix, it dynamically balances the multi-dimensional game relationships among the cost constraints of investors, the collaboration needs of scientific researchers, and the public’s expectations for open spaces. This transformation requires a paradigm-level reconstruction of professional ability standards: designers need to master new skills such as human-computer interaction protocol design, algorithm ethics review, and interdisciplinary knowledge distillation. Traditional aesthetic literacy and spatial operation ability are transformed into the ability to optimize the prompt engineering of intelligent systems and culturally decode the output results. The education system urgently needs to establish a cultivation framework for “critical technology thinking” so that designers can identify the value load in the algorithm black box and adhere to the leading position of humanistic spirit in human-machine collaboration. Conclusion Through theoretical construction and practical verification, this research reveals the innovative path and transformation boundaries of intelligent technology enabling environmental art design. At the theoretical level, the proposed “aesthetics-technology-human factors” ternary collaborative framework breaks through the binary opposition mindset of form and function in traditional design theories. By establishing a cross-modal attention mechanism and a dynamic weight allocation algorithm, it realizes the real-time interactive mapping of the cultural semantic network, engineering parameter system, and user behavior map, providing a computable cognitive interface for the digital translation of spatial aesthetics. This theoretical breakthrough not only explains the internal mechanism of creative emergence in the process of human-computer collaboration but also promotes the transformation of the environmental art discipline from an experience-dependent paradigm to a data-knowledge dual-driven paradigm by constructing a quantitative channel for design value transfer. In the practical dimension, model validation shows that this framework has significant technology spillover effects. Taking the renewal of smart city blocks as an example, by embedding real-time traffic flow and microclimate data, the system can dynamically adjust the height-width ratio and greening penetration rate of the street interface. While maintaining the integrity of the historical features, it can reduce the heat island intensity by 1.2°C. In the scenario of digital restoration of Dunhuang mural caves, the algorithm, through feature extraction of multi-spectral images and topological reconstruction of missing parts, increases the restoration efficiency of traditional craftsmen by 4 times, and the color restoration accuracy reaches 93%. These cross-field application cases confirm the tool value of the model in balancing cultural inheritance and technological innovation, providing an expandable technological foundation for new urbanization and the activation of cultural heritage. It is worth noting that the adaptability of the current technical system in cross-cultural contexts still has limitations. When dealing with aesthetic systems dominated by irregular forms (such as African tribal art or Austronesian architecture), algorithms are vulnerable to the cultural biases of training data, posing a risk of symbol misplacement. Future research needs to focus on breaking through the bottleneck of domain knowledge transfer under small-sample conditions. By constructing an interpretable style feature decoupling model, cross-cultural adaptation of aesthetic elements can be achieved. At the same time, develop a low-code interactive platform for designers, encapsulate underlying functions such as latent space exploration and parameter sensitivity analysis into visual modules, enabling intelligent design technology to break through the professional barriers of computer scientists and truly become a democratized tool for environmental art innovation. Declarations Data availability The datasets generated and/or analysed during the current study are not publicly available due to the proprietary nature of the intelligent design model and the core algorithms but are available from the corresponding author on reasonable request（ [email protected] ）. Author contributions Shaoqing Wang: Conceptualization, Methodology, Software, Formal Analysis, Writing – Original Draft Preparation. Jian Yao: Methodology, Validation, Investigation, Data Curation, Writing – Review & Editing. Shaoqing Wang proposed the research concept, developed the human-machine symbiotic intelligent design model, designed the generative adversarial network for spatial semantic analysis, and drafted the manuscript. Jian Yao contributed to the construction of the multi-objective dynamic optimization framework, implemented the augmented reality collaborative interface, conducted empirical validation, and performed data analysis. Both authors participated in the interpretation of results, manuscript revision, and approval of the final version. Funding The authors received no funding for this work. Competing interests The authors declare no competing interests. Ethics statement This study did not involve human participants requiring ethical approval beyond standard informed consent for data collection. All participant data were collected anonymously and used solely for academic research purposes. No personally identifiable information was retained. The research procedures were conducted in accordance with the ethical standards of Krirk University and the principles outlined in the Declaration of Helsinki. All authors confirm that there are no ethical conflicts related to this work. References Cai Qing, Liao Wenjie, Xue Hongjing, Huang Shengnan, Lu Xinzheng & Zhang Shuo.(2024). Application Method and Case Study of the Generative Intelligent Design Tool for Shear Wall Structure Schemes. Building Structure, 1 - 9. DOI: 10.19701/j.jzjg.LS243053. Cao Rongrong, Liu Lin, Yu Yandong & Wang Hailong.(2024). A review of research on large language models integrating knowledge graphs. Application Research of Computers, 1 - 14. DOI: 10.19734/j.issn.1001 - 3695.2024.12.0532. Li Hui, Wang Junyin, Cheng Yuanzhi, Liu Jian, Zhao Guowei & Chen Shuangmin. (2024). A 3D object detection method based on image semantic feature guidance and cross - modal fusion of point clouds. Journal of Computer - Aided Design & Computer Graphics (05), 734 - 749. Li Yubin. (2025). The cultural style, spatial narrative and image expression of documentaries on the Yellow River theme under the theory of cultural semiotics. Movie Review (05), 55 - 60. DOI: 10.16583/j.cnki.52 - 1014/j.2025.05.006. Liang Yinjie, Nan Xinyuan, Cai Xin, Li Yunpeng & Gou Haiguang.(2024). Bridge Defect Detection Based on Data Augmentation and Improved YOLOv8. Journal of Guangxi Normal University (Natural Science Edition), 1 - 16. DOI: 10.16088/j.issn.1001 - 6600.2024071003. Ma Yongjie & Chen Min. (2022). Dynamic Multi - objective Optimization Algorithm Based on Kalman Filter Prediction Strategy. Journal of Jilin University (Engineering and Technology Edition) (06), 1442 - 1458. DOI: 10.13229/j.cnki.jdxbgxb20210082. Nie Ying. (2024). Research on the Environmental Art Design Industry and Education in the AI Era. Art Design Research (05), 120 - 128. Pei Xuan. (2023). The Significance Structure Expression of “Sense of Place” and Spatial Aesthetics. Social Sciences in Yunnan (01), 153 - 163. Peng Lan. (2024). The Drift of Boundaries and the Guardianship of Subjectivity: Reflection on the Issue of Human Subjectivity under the Trend of Intelligent Content Production. Journal of Soochow University (Philosophy and Social Science Edition) (05), 159 - 170. DOI: 10.19563/j.cnki.sdzs.2024.05.015. Qin Junhui. (2020). Research on the environmental art design system based on virtual reality technology. Modern Electronics Technique (13), 92 - 94 + 98. DOI: 10.16652/j.issn.1004 - 373x.2020.13.022. Tang Yi. (2025). The Mechanism and Path of the Inheritance and Development of Excellent Traditional Chinese Culture from the Perspective of Design Aesthetics. Jiangsu Social Sciences (01), 202 - 210 + 244. DOI: 10.13858/j.cnki.cn32 - 1312/c.20250121.003. Wang Shishun. (2023). Research on the Creative Potential of Generative Adversarial Networks in the Field of Art Design. Packaging Engineering (S1), 24 - 28 + 41. DOI: 10.19554/j.cnki.1001 - 3563.2023.S1.004. Wang Wenzhong, Zhang Shusheng, Yu Suihuai & Chen Dengkai. (2021). Optimization of the cabin layout of submersibles based on deconstruction and reconstruction thinking. Computer Integrated Manufacturing Systems (07), 2045 - 2052. DOI: 10.13196/j.cims.2021.07.019. Wang Xingcheng, Wang Haoyu, Pan Xinyu, Tao Fang, Zheng Xingxing & Cao Shuang.(2025). Applications of Machine Learning in the Field of Environmental Microplastics: Classification and Identification, Quantitative Analysis, and Adsorption Performance Prediction. China Environmental Science, 1 - 15. DOI: 10.19674/j.cnki.issn1000 - 6923.20250507.002. Xiong Liang. (2024). Concerns, Optimization and Future Trends of the Educational Application of Affective Computing in the Era of Intelligent Technology. Open Education Research (06), 66 - 71. DOI: 10.13966/j.cnki.kfjyyj.2024.06.008. Xu Ming. (2025). The digital logic and its implementation path for improving the effectiveness of grass - roots governance. Inner Mongolia Social Sciences (03), 55 - 61. DOI: 10.14137/j.cnki.issn1003 - 5281.2025.03.008. Xu Muchuan & Wu Zijun. (2020). Research on the iteration and design of complex commercial complexes. Art Journal (02), 102 - 107. Xu Qianyao, Wu Qiong & Gong Wei. (2022). Analysis and Inspiration of Intelligent Design Methods Based on Aesthetic Computation. Decoration (09), 17 - 22. DOI: 10.16272/j.cnki.cn11 - 1392/j.2022.09.031. Zhang Delu.(2025). A Study on the Applicability of Social Semiotics in the Construction of Multimodal Discourse. Journal of Beijing International Studies University, 1 - 13. Zhang Jingyu, Liu Nianxiong, Wang Shanshan & Yan Shurui. (2022). Energy - saving Performance - oriented Intelligent Generation Design Method and Tool Platform for Residential Buildings. Architectural Journal (02), 22 - 27. DOI: 10.19819/j.cnki.ISSN0529 - 1399.202202004. Zhang Kaixuan, Cai Guoyong & Zhu Kunri. (2023). Visual Perception Recommendation System with Enhanced Image Aesthetic Information. Computer Science (S2), 285 - 292. Zhang Wenting, Liu Mandan, Zhu Baoxu & Xie Lizhi.(2024). Generation of 3D Facial Expression Animation Based on Interactive Evolutionary Algorithm. Journal of East China University of Science and Technology (Natural Science Edition), 1 - 15. DOI: 10.14135/j.cnki.1006 - 3080.20240831001. Zhou Xiang, Cui Lichen, Chen Suqing & Tan Zimei. (2024). Visual Salience Analysis of Imagery Elements in Historical Blocks Based on Eye - tracking: Taking Yongqingfang in Guangzhou as an Example. Modern Urban Research (10), 101 - 107. Zuo Wenjin, He Xiaogang, Wen Chuanzhen & Liu Lijun. (2022). Research on the construction of the evaluation index system for the quality of big data resources - Analysis of library big data from the perspective of user perception. Price:Theory & Practice (08), 55 - 58. DOI: 10.19851/j.cnki.CN11 - 1010/F.2022.08.445. 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-7464844","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":513951098,"identity":"b69dc77a-2195-44f6-8107-16b90006fe95","order_by":0,"name":"Shaoqing Wang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3ElEQVRIiWNgGAWjYDACCRBhwMDAz8zY+CChooYELZLtzM0GD84cI1YLSNd59jbJhy3MhHXIz25++OBNwR27hsOMbRWJDWwM/O3dCXi1GNw5Zmw4x+BZcmMzY9uNxB0yDBJnzm7Ar0UiwUyax+BwMjMzSMsZNqBILn4t8jPSv4G1sAG1FCS2MRPWwnAjB2yLHQ9QCwNRWgxu5BQD/XI4QYKZsVki4cwxHoJ+ATps44M3fw7b258//vDjj4oaOf72XgIOAwEeBobEBgSbGABUZk+cylEwCkbBKBiRAADIf0kVcozXQQAAAABJRU5ErkJggg==","orcid":"","institution":"Krirk University","correspondingAuthor":true,"prefix":"","firstName":"Shaoqing","middleName":"","lastName":"Wang","suffix":""},{"id":513951100,"identity":"15e50b7f-005b-4261-9038-ae12d9190e71","order_by":1,"name":"Jian Yao","email":"","orcid":"","institution":"Krirk University","correspondingAuthor":false,"prefix":"","firstName":"Jian","middleName":"","lastName":"Yao","suffix":""}],"badges":[],"createdAt":"2025-08-26 16:23:09","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7464844/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7464844/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":91963785,"identity":"829f4e39-954b-4c9c-b903-19ed29f0d22d","added_by":"auto","created_at":"2025-09-23 08:10:45","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":232595,"visible":true,"origin":"","legend":"","description":"","filename":"10002Figure.docx","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/6f6f096be05f9da29099cd18.docx"},{"id":91847719,"identity":"230a8f60-4456-48a1-ba3d-fc8458e43f85","added_by":"auto","created_at":"2025-09-22 10:27:01","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":276871,"visible":true,"origin":"","legend":"","description":"","filename":"10002S.W.IntelligentDesignEmpowersSpatialAestheticsConstructionofaManMachineCollaborativeInnovationModelforEnvironmentalArtandCaseVerification32.docx","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/eff8ef464003bacf997fe89f.docx"},{"id":91846301,"identity":"a8b00abb-64a0-40fb-a473-a450eb0ea2b7","added_by":"auto","created_at":"2025-09-22 10:19:01","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":16528,"visible":true,"origin":"","legend":"","description":"","filename":"10002Table.docx","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/860f4f283588b496127147db.docx"},{"id":91847723,"identity":"5fd4df67-8112-442b-9730-a34e5390ae67","added_by":"auto","created_at":"2025-09-22 10:27:01","extension":"json","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4536,"visible":true,"origin":"","legend":"","description":"","filename":"0bcaf2ece9f149a7a6f903d6b10e0139.json","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/149e5a51d2fe1dc2bbf59e92.json"},{"id":91846311,"identity":"5b8f9ee9-5421-4c6e-93e8-0ba8e998cf19","added_by":"auto","created_at":"2025-09-22 10:19:01","extension":"xml","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":120787,"visible":true,"origin":"","legend":"","description":"","filename":"0bcaf2ece9f149a7a6f903d6b10e01391enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/1f8e65a83db7f9a67f73743b.xml"},{"id":91848022,"identity":"9b8a3280-67f3-41bc-b40a-01e1797715e4","added_by":"auto","created_at":"2025-09-22 10:35:01","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":55003,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/e62fd4141867c4f266154086.png"},{"id":91848474,"identity":"442d0901-19f6-4390-a03f-e3705a5a1e99","added_by":"auto","created_at":"2025-09-22 10:43:01","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":129929,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/0a2833ea36753657aaf8b874.png"},{"id":91848699,"identity":"2c2d0418-1614-49e1-ad39-9e8a9e038bff","added_by":"auto","created_at":"2025-09-22 10:51:01","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":29740,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/bd1a7f8dbeabf5df34d184ea.png"},{"id":91846310,"identity":"880a94c8-a9c1-42a0-b85b-86b94441cadc","added_by":"auto","created_at":"2025-09-22 10:19:01","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":55003,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/9739c9c7153b3983425e6b08.png"},{"id":91848017,"identity":"bbbc50c1-5458-4f5f-ae2f-51eaf8a02777","added_by":"auto","created_at":"2025-09-22 10:35:01","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":129929,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/4ffd1b68674c76e85c67efdf.png"},{"id":91846307,"identity":"bfbbbc54-af84-4502-9912-504c154682aa","added_by":"auto","created_at":"2025-09-22 10:19:01","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":29740,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/68a91def0fb9f9721fcf9072.png"},{"id":91846316,"identity":"2da58d28-9cf2-41a2-b718-06f25c0123a5","added_by":"auto","created_at":"2025-09-22 10:19:01","extension":"wmf","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":184,"visible":true,"origin":"","legend":"","description":"","filename":"image1.wmf","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/3b4bbffcdd8f134a1e3b8bd7.wmf"},{"id":91847725,"identity":"99b5dac1-5ff1-44d1-81e5-3a345fde089e","added_by":"auto","created_at":"2025-09-22 10:27:01","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10252,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/7f20b519b833e1715c33d90e.png"},{"id":91848020,"identity":"c5620302-582a-4b98-b3ed-2a7ec1cfc429","added_by":"auto","created_at":"2025-09-22 10:35:01","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":16522,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/5c4e1d4e112bc01c24a67093.png"},{"id":91846317,"identity":"86714b7d-9ec9-42ff-9254-ebdbd5401db5","added_by":"auto","created_at":"2025-09-22 10:19:01","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9109,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/f13de366d09eb491ce2751c1.png"},{"id":91846312,"identity":"252d8536-6c35-4365-9601-01d3a7dc9d94","added_by":"auto","created_at":"2025-09-22 10:19:01","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10252,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/6a225be8d31bb7b5a4ea1603.png"},{"id":91847727,"identity":"b229a8c2-80da-4025-98dc-ca0b9acfaa75","added_by":"auto","created_at":"2025-09-22 10:27:01","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":16522,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/099680d05075d3bc17ccda5b.png"},{"id":91847730,"identity":"4448d564-83fa-4b2e-ad7f-3a4b71d06f8f","added_by":"auto","created_at":"2025-09-22 10:27:01","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9109,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/22ae14f66791cf37150784bf.png"},{"id":91846320,"identity":"fe785dc8-fce8-49c4-9784-045d71f8bce4","added_by":"auto","created_at":"2025-09-22 10:19:01","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":136,"visible":true,"origin":"","legend":"","description":"","filename":"Onlineimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/b453d9d5fe4561cf143542be.png"},{"id":91846318,"identity":"03f083ef-ec60-442b-b986-4fe28aba4265","added_by":"auto","created_at":"2025-09-22 10:19:01","extension":"xml","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":119375,"visible":true,"origin":"","legend":"","description":"","filename":"0bcaf2ece9f149a7a6f903d6b10e01391structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/b5362decb155af7dbb44f04d.xml"},{"id":91846322,"identity":"c1f75f97-66db-411c-a07f-2de20a851d47","added_by":"auto","created_at":"2025-09-22 10:19:01","extension":"html","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":126747,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/6a1493e1a8ef0d7d6e464619.html"},{"id":91846297,"identity":"9a37eab5-26a7-4112-86f7-cc8ee06c4cba","added_by":"auto","created_at":"2025-09-22 10:19:01","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":42355,"visible":true,"origin":"","legend":"\u003cp\u003eComparative analysis of design efficiency\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/ae326df467883316cc699eb5.png"},{"id":91846299,"identity":"24e52fff-e5a3-4e57-8609-9b91c54eef12","added_by":"auto","created_at":"2025-09-22 10:19:01","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":75339,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of optimized parameters for the cultural exhibition hall plan\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/5dddc236afbf0d0c7f937887.png"},{"id":91846298,"identity":"693d19d6-324f-4aee-ba51-4ddc486d7786","added_by":"auto","created_at":"2025-09-22 10:19:01","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":17976,"visible":true,"origin":"","legend":"\u003cp\u003eCorrelation analysis between expert scores and algorithm predicted values\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/2daa31c112eed6bae2fd42a7.png"},{"id":104423525,"identity":"3e031f59-606b-4910-81f3-d3fbd3f43c66","added_by":"auto","created_at":"2026-03-11 14:13:17","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":828951,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7464844/v1/ac6e36e6-4f01-425a-9288-96b7fa93cc6a.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Intelligent design empowers spatial aesthetics: construction of a man-machine collaborative innovation model for environmental art and case verification","fulltext":[{"header":"Introduction","content":"\u003cp\u003e\u003cstrong\u003eResearch background\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn recent years, the deep penetration of artificial intelligence and big data technologies is triggering a paradigm reconstruction in the field of environmental art design\u003csup\u003e1\u003c/sup\u003e. Intelligent algorithms represented by generative design and deep learning are gradually breaking through the capacity boundaries of traditional tools in spatial form deduction and aesthetic law analysis. Through the feature deconstruction and non-linear correlation mining of a vast number of cases, they have achieved a leap from static plan output to dynamic creative generation. Especially in the scenario of functional-aesthetic coupling optimization in complex spaces, algorithm-driven high-dimensional parametric modeling can simultaneously handle lighting relationships, material properties, and ergonomic constraints, significantly improving the accuracy and systematicness of design iteration and providing quantifiable technical support for multi-objective decision-making.\u003c/p\u003e\n\u003cp\u003eHowever, the field of environmental art design still faces the dual challenges of experience dependence and technical adaptability. In current industry practices, designers\u0026rsquo; shaping of spatial aesthetics is mostly based on individual aesthetic qualities and the accumulation of project experience. The creative mode dominated by subjective judgment not only leads to a long scheme generation cycle and low resource allocation efficiency, but also makes it difficult to meet the high-level requirements for dynamic adaptability in emerging scenarios such as urban renewal and smart buildings. At the same time, the professional barriers among architecture, landscape, and interior design in the engineering implementation stage make it difficult to effectively integrate interdisciplinary data, often resulting in the separation of aesthetic expression and functional logic, and exacerbating the construction rework rate and energy consumption.\u003c/p\u003e\n\u003cp\u003eAgainst this backdrop, there is an urgent need to construct a collaborative framework of engineering rationality and artistic sensibility for the innovative path of spatial aesthetics. On the one hand, as a material carrier, space needs to meet the requirements of physical properties such as structural stability and environmental compatibility. On the other hand, as a cultural medium, it must also fulfill the social functions of carrying emotional resonance and transmitting aesthetic values. The one-way thinking of \u0026ldquo;technology first\u0026rdquo; or \u0026ldquo;form supremacy\u0026rdquo; in traditional design methods can no longer meet the demands of the era that emphasizes both people-oriented orientation and sustainable development. By introducing a human-machine collaborative mechanism, the intelligent system can expand the creative boundaries relying on its data mining capabilities, while designers can make aesthetic adjustments and value judgments on the algorithm output with critical thinking\u003csup\u003e2\u003c/sup\u003e. The in-depth interaction between the two will drive the design process to shift from linear progression to closed-loop optimization, ultimately achieving an organic unity of technological empowerment and humanistic care.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResearch significance\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe integrated innovation of intelligent technology and environmental art design is essentially a dual expansion of the ontology and methodology of design. From a theoretical dimension, transforming the formal rules and emotional experiences of spatial aesthetics into a computable and optimizable parameter system\u003csup\u003e3 \u003c/sup\u003enot only breaks through the cognitive paradigm dominated by qualitative descriptions in traditional design research, but also provides a systematic framework for interdisciplinary knowledge integration by establishing a mapping relationship of \u0026ldquo;aesthetic features - algorithm logic - human factor feedback\u0026rdquo;. This aesthetic computing model based on data density and algorithm complexity can effectively decode the implicit laws in the evolution of historical styles and capture the dynamic changes in the aesthetic preferences of contemporary users. Thus, it builds a two-way verification channel between quantitative analysis and qualitative judgment, promoting the paradigm innovation of design theory from empirical induction to deductive reasoning.\u003c/p\u003e\n\u003cp\u003eIn the practical dimension, the collaborative design mechanism empowered by algorithms is reconstructing the value chain of environmental art creation. By embedding multi-objective optimization algorithms and real-time rendering technologies, the intelligent system can quickly generate candidate solutions that balance functional rationality and visual expressiveness\u003csup\u003e4\u003c/sup\u003e. It also incorporates dynamic variables such as user behavior data and environmental sensor information into the iterative cycle, significantly enhancing the adaptability of design solutions to complex scenarios. This data-driven decision-making model not only reduces the time and resource consumption brought about by the traditional trial-and-error method but also breaks the information silos among sub-fields such as architecture, landscape, and interior design by establishing a cross-professional collaboration platform. As a result, the creation of spatial aesthetics has shifted from discrete fragmented operations to integrated innovation of full-life-cycle management. Especially in emerging fields such as smart city construction and digital protection of cultural heritage, the human-machine collaborative model provides a scalable solution for balancing technological innovation and cultural inheritance\u003csup\u003e5\u003c/sup\u003e. Its dynamic response characteristics are more in line with the requirements of the times for intensive utilization of resources under the goal of sustainable development. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResearch questions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe construction of the collaborative paradigm between intelligent systems and human designers essentially aims to address the dialectical relationship between machine logic and humanistic values in environmental art creation. The primary challenge lies in how to design interaction rules and responsibility boundaries to make the efficient computing ability of algorithms complement rather than compete with the aesthetic judgment of designers. This requires the establishment of a phased and hierarchical collaborative mechanism: in the concept generation stage, the intelligent system needs to break through the stylistic limitations of traditional parametric tools and capture unstructured aesthetic features through generative adversarial networks\u003csup\u003e6\u003c/sup\u003e; while in the scheme refinement stage, designers need to take the lead in interpreting cultural symbols and constructing spatial narratives to ensure that the candidate schemes output by the algorithm have adaptable creative extensibility. The collaborative efficiency of the two not only depends on the friendliness of the technical interface but also relies on the accuracy of the algorithmic translation of the tacit knowledge of design thinking.\u003c/p\u003e\n\u003cp\u003eFurthermore, the evaluation of the improvement effect of spatial aesthetic quality needs to break through the ambiguity of traditional subjective evaluation and construct a composite index system that integrates objective data and humanistic criteria. This involves the quantitative connection of three dimensions: First, through the style classification model and emotion computing technology\u003csup\u003e7\u003c/sup\u003e, the aesthetic features are deconstructed into measurable parameters such as color contrast and morphological complexity entropy. Second, an expert experience database is established, and the analytic hierarchy process is used to determine the weight distribution of each aesthetic element. Third, eye-tracking and physiological signal monitoring are introduced to capture users\u0026rsquo; subconscious feedback on the spatial atmosphere\u003csup\u003e8\u003c/sup\u003e. Only by achieving the multi-source alignment of algorithm predicted values, expert scores, and user perception data can an intelligent design evaluation benchmark with industry consensus be formed. \u003c/p\u003e\n\u003cp\u003eIn addition, when the human-machine collaborative model is applied to complex scenarios such as the renewal of urban complexes and historical blocks, its robustness needs to be ensured through a multi-scale verification framework. At the technical level, it is necessary to test the generalization ability of the algorithm in data-sparse scenarios. For example, the style characteristics of existing projects can be adapted to heterogeneous cultural contexts through transfer learning. At the application level, it is necessary to evaluate the response sensitivity of the model to dynamic variables, including real-world constraints such as sudden changes in functional requirements and aesthetic conflicts among multiple stakeholders. Only by ensuring that the algorithm strikes a balance between stability and adaptability can the potential of the intelligent design system to transform from an experimental tool to an engineered platform be confirmed. \u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResearch objectives\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo achieve the intelligent transformation of the environmental art design paradigm, it is necessary to establish the research objective orientation from the dual perspectives of technological innovation and value reconstruction. The primary task is to construct the technical ontology of the man-machine collaborative innovation model. By deconstructing the cognitive decision-making chain in the design process, the role division of the intelligent system and designers at different stages can be clarified. At the technical path level, it is necessary to integrate the style transfer ability of the generative adversarial network, the constraint solving efficiency of the multi-objective optimization algorithm, and the immersive tuning function of virtual reality interaction to form a closed-loop architecture of \u0026ldquo;data perception - plan generation - dynamic optimization\u0026rdquo;. The design of interaction rules focuses on the dynamic allocation mechanism of creative control rights. For example, the algorithm is given higher autonomy during the spatial form topological optimization stage, while the priority of manual intervention is strengthened in the cultural semantics injection link, so as to achieve the gradient integration of machine logic and human creativity.\u003c/p\u003e\n\u003cp\u003eOn this basis, the research objective further focuses on the verification of the practical effectiveness of the model. A scientific evaluation system needs to be established through case-based empirical studies in multiple scenarios and at multiple scales. Regarding the aesthetic expression dimension, two types of projects with vastly different styles, namely the revitalization of historical blocks and futuristic pavilions, are selected. The differences in the algorithm\u0026rsquo;s expressiveness in cultural symbol extraction and avant-garde form interpretation are comparatively analyzed. At the functional adaptation level, the real-time feedback of sensor data on the spatial utilization efficiency of commercial complexes is used to verify the response accuracy of the intelligent system to dynamic variables such as passenger flow density and energy consumption indicators. For the verification of user experience, eye movement trajectory heatmaps and emotional semantic analysis are relied upon to quantitatively evaluate the improvement amplitude of the human-machine collaboration scheme in terms of spatial narrative coherence and emotional resonance intensity. This full-life-cycle empirical framework will reveal the transformation boundaries and adaptation conditions of intelligent design models from theoretical construction to engineering implementation.\u003c/p\u003e"},{"header":"Literature review","content":"\u003cp\u003e\u003cstrong\u003eResearch progress of intelligent design\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe evolution of intelligent design technology is profoundly reconstructing the methodological system of architectural form generation and aesthetic analysis\u003csup\u003e9\u003c/sup\u003e. Generative design, by embedding genetic algorithms and topological optimization principles, transforms the traditional form creation relying on empirical intuition into a multi-objective-driven parametric solution process. Its core advantage lies in simultaneously handling the complex coupling relationships among structural performance, material efficiency, and spatial experience. Taking large public buildings as an example, through iterative calculations, the algorithm can achieve a dynamic balance between load-bearing efficiency and curved surface aesthetics, generating irregular structure schemes that are both mechanically reasonable and visually appealing. This design paradigm of automatically optimizing based on constraints significantly improves the solution efficiency of complex engineering problems and the diversity of design schemes.\u003c/p\u003e\n\u003cp\u003eMeanwhile, the intervention of deep learning technology has further expanded the aesthetic analysis dimension of intelligent design. The theoretical foundation of this technological evolution can be traced back to the paradigm breakthrough of aesthetic computing - it quantifies formal rules and emotional experiences into high-dimensional computable parameters, constructs an interactive interface between artistic creation and algorithmic logic, and promotes the transformation of environmental art design from an experience-driven paradigm to a data-knowledge dual-engine supported one. By extracting the style features of the architectural historical image library through convolutional neural networks\u003csup\u003e10\u003c/sup\u003e, the system can establish quantitative representation models of aesthetic laws such as the vertical rhythm of Gothic style and the simple proportion of modernism, and achieve the contemporary translation of traditional vocabulary with the help of style transfer algorithms. This data-driven feature reconstruction ability enables intelligent systems not only to imitate established styles, but also to generate novel forms that break through the boundaries of experience through exploration in the latent space. For example, regional cultural symbols can be deconstructed into topologically variable basic units and then recombined into innovative forms that meet contemporary functional requirements. However, the existing technology still faces a structural mismatch between the generated results and humanistic aesthetic standards. Especially in the deep aesthetic dimensions such as cultural metaphor conveyance and emotional resonance intensity, algorithmic logic is still difficult to completely replace the critical thinking of human designers.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTechnological challenges of spatial aesthetics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe penetration of intelligent design technology into the field of spatial aesthetics has always faced the structural tension between machine logic and humanistic values. Although the algorithm-generated solutions perform excellently in terms of morphological complexity and engineering rationality, there remains an essential contradiction in the degree of fit between their aesthetic output and the common human aesthetic consensus. This contradiction stems from the context-dependence of aesthetic judgment: The style association models established by algorithms through massive data training often simplify cultural symbols into probability distributions of visual features, but it is difficult to capture unstructured elements such as the sense of ritual and collective memory in the local context. For example, in the scenario of historical block renewal, the generated solutions may accurately reproduce the proportional scale of traditional architecture, but lose the time narrative carried by the aging traces of materials, resulting in the spatial experience falling into \u0026ldquo;precise paleness\u0026rdquo;. More importantly, the current aesthetic evaluation datasets mostly originate from the Western modernist paradigm. This cultural bias may cause the algorithm to overemphasize geometric order when generating irregular forms, suppressing the expression of non-linear characteristics such as \u0026ldquo;blank space\u0026rdquo; and \u0026ldquo;artistic conception\u0026rdquo; in Eastern aesthetics.\u003c/p\u003e\n\u003cp\u003eWhen dealing with dynamic spatial requirements, the rigid characteristics of the technical framework of parametric tools are becoming increasingly prominent\u003csup\u003e11\u003c/sup\u003e. Although topological optimization algorithms can efficiently handle the generation of forms under static constraints, when faced with dynamic variables such as sudden changes in users\u0026rsquo; behavior patterns and real-time changes in environmental lighting, the existing toolchains often exhibit system inertia. This limitation is essentially a contradiction between the \u0026ldquo;priori setting\u0026rdquo; methodology of parametric modeling and the \u0026ldquo;emergent characteristics\u0026rdquo; of the real world: designers need to pre-define all the parameters that may affect the spatial form and their weight of influence, while the dynamic interactions in reality often produce associative effects beyond the pre-set scope. Taking urban public spaces as an example, real-time data such as pedestrian flow density and types of social activities should drive the adaptive adjustment of the spatial interface. However, due to the lack of an online learning mechanism in traditional parametric systems, it is difficult to convert sensor data streams into form adjustment instructions, resulting in dynamic design remaining at the conceptual stage. This tool limitation not only restricts the scene response ability of spatial aesthetics but also hinders the evolution of human-computer collaborative design towards full-life-cycle management.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eParadigm of human-machine collaborative design\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe evolution of the human-machine collaborative design paradigm is redefining the boundaries and implementation paths of creativity\u003csup\u003e12\u003c/sup\u003e. Within the theoretical framework of role division of labor, the collaboration between designers and intelligent systems is not simply a task segmentation, but a differential empowerment based on cognitive advantages: designers take the lead in value judgment and cultural decoding, and are responsible for setting aesthetic goals and ethical constraints; intelligent systems, on the other hand, give full play to their computational advantages in hyperparameter space exploration and multi-objective optimization, and transform abstract aesthetic demands into executable morphological strategies. This collaborative model of \u0026ldquo;humans setting the direction - machines expanding possibilities\u0026rdquo; is particularly evident in the digital restoration of historical buildings. Designers endow the cultural context of space restoration through semantic networks, and algorithms generate technical solutions that conform to structural specifications and style continuity based on this. The two continuously calibrate the creative trajectory through a real-time feedback loop to ensure that technical rationality always serves the expression of humanistic values.\u003c/p\u003e\n\u003cp\u003eThe introduction of interactive evolutionary algorithms provides a dynamic creative tuning mechanism for human-computer collaboration\u003csup\u003e13\u003c/sup\u003e. This technology converts the designer\u0026rsquo;s subjective evaluation into a fitness function, enabling the algorithm to continuously capture human preference characteristics during the iterative process, and thus achieve the progressive optimization of the generation direction. Taking commercial space design as an example, the designer\u0026rsquo;s aesthetic preference scores for the initial set of design schemes are encoded as the retention probabilities of morphological genes, driving the algorithm to strengthen key features such as curve softening and material contrast in subsequent iterations, while automatically filtering out structural proposals that violate ergonomics. This semi-automated evolutionary process not only retains the human\u0026rsquo;s ultimate right to decide on aesthetic quality but also breaks through the mindset of traditional design processes through the machine\u0026rsquo;s parallel computing capabilities. More importantly, the interactive evolutionary algorithm constructs an interpretable creative evolution path, enabling designers to trace the generation logic of specific aesthetic characteristics, thereby establishing a verifiable cognitive bridge between perceptual judgment and algorithmic decision-making.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResearch gaps\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWhile promoting the innovation of technical tools in current intelligent design research, there are still crucial discontinuities in its methodological system. The primary limitation lies in the disconnection between the aesthetic evaluation dimension and the algorithm generation process: most models simplify aesthetics to the optimization of the probability distribution of morphological parameters, but fail to establish a quantitative index system that is deeply associated with regional culture and emotional experience. This technological tendency results in a paradox where the generated solutions often exhibit \u0026ldquo;formal correctness\u0026rdquo; but \u0026ldquo;meaning vacuity\u0026rdquo;. For example, although algorithms can accurately fit the proportional characteristics of new Chinese-style architecture, they are unable to decode the metaphorical relationship between the upturned angle of the eaves and the spirit of the place, and it is even more difficult to quantitatively evaluate the intensity of cultural identity that users experience when they are in the space. The lack of such an evaluation system turns the intelligent design system into a style collage tool rather than a true aesthetic innovation engine.\u003c/p\u003e\n\u003cp\u003eA more profound limitation lies in the one-sidedness of the case verification paradigm. Most existing studies focus on proving the superiority of algorithms in technical indicators such as generation speed and structural rationality, but rarely construct an interdisciplinary evaluation framework that encompasses art value criticism and social acceptance analysis. Taking parametric public art installations as an example, research literature generally emphasizes the topological optimization efficiency of form generation algorithms, but ignores core art value dimensions such as the dialogue ability between the installation and the urban context and public participation. This verification bias may mislead the direction of technology research and development, causing intelligent design to fall into the dilemma of being \u0026ldquo;technically feasible\u0026rdquo; but \u0026ldquo;culturally silent\u0026rdquo;. Even more severe is that when algorithm-generated solutions involve sensitive scenarios such as the renovation of historical blocks, the lack of a social acceptance evaluation mechanism will amplify the technical ethics risks, leading to a cultural identity crisis for the design solutions during the implementation phase.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cp\u003e\u003cstrong\u003eData source and processing\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe acquisition and processing mechanism of research data is a fundamental task to support the construction of the human-machine collaborative model. The construction of the spatial aesthetics feature database follows the principles of scene taxonomy, covering 10 types of typical environments such as commercial complexes, cultural exhibition halls, and historical blocks. Through 3D laser scanning and high-precision texture mapping technologies, more than 5,000 design schemes are transformed into computable data entities. Each case is subject to multi-layer semantic annotation: in the style dimension, a convolutional neural network is used to automatically identify the regional feature index\u003csup\u003e14\u003c/sup\u003e. The material properties are double-verified through spectral reflectance and tactile simulation parameters, and the lighting data at different times is reconstructed for the light and shadow relationship relying on the radiosity algorithm. To ensure the consistency of the annotation logic, an expert-algorithm collaborative verification mechanism is adopted. Scholars in the fields of architectural history and environmental psychology calibrate the feature labels in the cultural context to eliminate the meaning distortion caused by simple visual feature matching.\u003c/p\u003e\n\u003cp\u003eThe quantitative collection of users\u0026rsquo; aesthetic preferences is achieved relying on the multi-modal perception technology system\u003csup\u003e15\u003c/sup\u003e. The eye-tracking system records the fixation point heat map of observers during virtual space roaming at a sampling rate of 250Hz, and analyzes the distribution of visual attractiveness through the dwell time of gaze and the complexity of saccade paths. At the same time, a wrist-worn physiological sensor synchronously collects galvanic skin responses and heart rate variability to establish an association model between spatial atmosphere stimuli and emotional arousal levels. To improve the ecological validity of the data, the experimental scenario uses augmented reality technology to overlay real environmental lighting and acoustic parameters, ensuring that the feedback from participants reflects the real spatial experience. After time series alignment and outlier removal of the original data, it is transformed into feature vectors with spatial grid encoding, and finally a multi-dimensional evaluation matrix that couples style preferences, functional requirements, and emotional feedback is formed, providing a training benchmark for the adaptive optimization of the intelligent system.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTechnical tools\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe construction of the technical toolchain follows the collaborative logic of \u0026ldquo;generation - optimization - interaction\u0026rdquo;, aiming to open up the value transfer channel from algorithmic computing to manual decision - making. The generation module adopts the improved StyleGAN3 architecture\u003csup\u003e16\u003c/sup\u003e. Through adaptive artifact suppression and latent space disentanglement technologies, it significantly improves the output quality of irregular building forms. During the training phase, this module injects knowledge in the field of architecture, deconstructs the historical case library into style gene segments, enabling the generator to recombine innovative solutions that not only conform to formal aesthetics but also avoid copyright disputes. Compared with traditional generative adversarial networks, the improved architecture increases the diversity of solutions by 37% while maintaining style continuity, effectively alleviating the constraints of mode collapse on creative exploration.\u003c/p\u003e\n\u003cp\u003eOn this basis, the Multi-Objective Genetic Algorithm (MOGA) constructs a dynamic balance mechanism between functionality and aesthetics\u003csup\u003e17\u003c/sup\u003e. The algorithm encodes indicators such as structural rationality, construction cost, and visual attractiveness into a multi-dimensional fitness function. Through the elitist retention strategy and non-dominated sorting, it screens out the optimal solution set on the Pareto front. In particular, for the conflicting constraints in spatial design, the algorithm introduces a dynamic weight adjustment mechanism. When the user feedback data shows a significant deviation during the scheme refinement stage, it automatically reduces the priority of the morphological complexity index to ensure that the optimization direction always targets the core requirements. This flexible optimization strategy increases the efficiency of scheme iteration by 58% compared to traditional methods, while maintaining the baseline standard of aesthetic quality.\u003c/p\u003e\n\u003cp\u003eFurthermore, implement embodied operations of human-machine collaboration through a virtual reality interactive platform\u003csup\u003e18\u003c/sup\u003e. This platform integrates a ray tracing rendering engine and a physical simulation system, which can real-time present the material reflection characteristics and spatial scale perception of design schemes under real lighting conditions. Designers directly grasp and stretch the key parameter nodes of virtual models through the gesture control interface, and the system simultaneously updates morphological indicators and performance data such as acoustics and thermal engineering. This what-you-see-is-what-you-get interactive mode not only reduces the time-consuming of scheme adjustment to 1/5 of that of traditional CAD tools, but also constructs an interpretable creative evolution path through the parameter operation history tracing function. The platform is specifically equipped with a collaborative mode, allowing the algorithm to continuously provide alternative scheme branches during the designer\u0026rsquo;s modification process, thus forming a two-way inspiring creative dialogue.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eModel construction\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe technical architecture of the human-machine collaborative innovation model follows the construction logic of \u0026ldquo;data-driven - intelligent emergence - value closed-loop\u0026rdquo;, and realizes the digital translation of design thinking through a hierarchical and progressive structure. At the data layer, the multi-modal feature extraction engine deconstructs spatial aesthetic elements into style vectors, material fingerprints, and light and shadow matrices. At the same time, it constructs a knowledge graph in a cross-cultural context to form a quantifiable and comparable design semantic network\u003csup\u003e19\u003c/sup\u003e. The algorithm layer adopts a cross-modal attention mechanism to achieve the dynamic alignment of text requirement descriptions, 3D point cloud data, and user physiological feedback signals\u003csup\u003e20\u003c/sup\u003e. Through the gated recurrent unit, it controls the feature weight distribution in different design stages to ensure that functional constraints and aesthetic goals achieve the Pareto frontier optimal solution in the parameter space. The application layer deploys a lightweight inference engine to convert the high-dimensional algorithm output into interactive BIM model components in real time, and establishes a two-way feedback channel to capture the designer\u0026rsquo;s adjustment intentions, forming a continuously evolving design knowledge base.\u003c/p\u003e\n\u003cp\u003eThe core of the collaborative mechanism lies in the formulation of a dynamic rights and responsibilities allocation strategy\u003csup\u003e21\u003c/sup\u003e. At the initial stage of design, the system generates a set of candidate solutions that go beyond the constraints of conventional experience through potential space exploration. It uses style interpolation technology to present a continuous spectrum from conservative to radical, inspiring designers to break through mental stereotypes. When the design plan enters the deepening stage, the collaborative rules automatically switch to the \u0026ldquo;human-dominated, machine-assisted\u0026rdquo; mode: the algorithm turns to conducting sensitivity analysis of detailed parameters, predicting the impact gradient of material replacement or morphological fine-tuning on the overall aesthetic evaluation, while designers focus on the topological adaptation of cultural symbols and the construction of spatial narratives\u003csup\u003e22\u003c/sup\u003e. In the final decision-making process, the built-in ethical review module of the system will detect whether the design plan has the risk of cultural stereotypes or characteristics of group exclusion. At the same time, it provides an interpretability report to clarify the algorithmic basis for key design decisions, ensuring the coexistence and evolution of technical rationality and humanistic values in spatial construction.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCase verification design\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe case verification design follows the principle of \u0026ldquo;scenario differentiation - indicator multi - dimensionalization - process traceability\u0026rdquo;. The universality of the model\u0026rsquo;s effectiveness is verified by ensuring full coverage of typical spatial types. The experimental objects focus on three types of scenarios with representative challenges: Urban public spaces mainly verify the model\u0026rsquo;s response ability to social inclusiveness and functional complexity; commercial complexes test the balance accuracy of human - machine collaboration in terms of economic benefits and user experience\u003csup\u003e23\u003c/sup\u003e; cultural exhibition halls mainly examine the possible boundaries of the integration of traditional symbol translation and avant - garde technologies. To ensure the ecological validity of the experiment, each type of scenario includes two project types: new construction and renovation. And the geographical distribution of cases is limited to cover different climate zones and cultural circles, so as to test the adaptability of the model under cross - environmental constraints.\u003c/p\u003e\n\u003cp\u003eThe construction of the evaluation index system breaks through the single dimension of the traditional technology-oriented approach\u003csup\u003e24\u003c/sup\u003e. Innovatively, it incorporates efficiency gain, professional consensus, and user perception into a unified analysis framework. The design cycle shortening rate calculates the total time compression ratio and the efficiency improvement value of key decision points by comparing the time-consuming differences between the human-machine collaborative model and the traditional process in key stages such as concept generation and scheme deepening\u003csup\u003e25\u003c/sup\u003e. For the consistency of expert scores, an interdisciplinary review group is formed. The double-blind test method is used to independently score indicators such as the cultural fit and technical rationality of the scheme. The Cohen\u0026rsquo;s Kappa coefficient is used to quantify the judgment consensus among architectural historians, structural engineers, and art curators. A threshold of \u0026ge; 0.75 is set as the baseline for the reliability of aesthetic evaluation. The user satisfaction survey abandons the single overall score. The Likert 5-point scale is used to collect data on sub-dimensions such as spatial readability, emotional resonance, and functional convenience in a hierarchical manner. The principal component analysis is used to extract the latent variables affecting satisfaction and establish a mapping relationship with the algorithm predicted values. All evaluation processes record operation logs and decision-making paths to ensure that the verification results are reproducible and auditable.\u003c/p\u003e\n\u003cp\u003eTo quantify the intervention effect of the intelligent design model, Cohen\u0026rsquo;s d value is used for effect size analysis. Its calculation formula is:\u003c/p\u003e\n\u003cp\u003e\u003cimg src=\"https://myfiles.space/user_files/58895_8739fc6c57c1c19a/58895_custom_files/img1758288423.png\" width=\"475\" height=\"95\"\u003e\u003c/p\u003e\n\u003cp\u003eThe calculation method of the combined standard deviation SDmerge\u0026nbsp;is as follows:\u003c/p\u003e\n\u003cp\u003eIn the formula: M experimental group represents the mean of the data of the experimental group; M control group represents the mean of the data of the control group; SD1 represents the standard deviation of the data of the experimental group; SD2 represents the standard deviation of the data of the control group; n1 represents the sample size of the experimental group; n2 represents the sample size of the control group. The interpretation criteria for the effect size refer to Cohen\u0026rsquo;s guidelines: small (d = 0.2), medium (d = 0.5), large (d = 0.8).\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eModel performance\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe comparative test data based on three typical scenarios show that the human-machine collaborative model has significantly improved in terms of design efficiency and aesthetic quality. In the efficiency dimension, by recording the time cost and the frequency of plan iteration in the conceptual design stage, it is found that compared with the traditional workflow, the intelligent system increases the generation efficiency of creative plans by 42% (the average value drops from 18.3 hours per plan to 10.6 hours), and reduces the number of design iterations by 60% (the average value is compressed from 7.2 times to 2.9 times). This gain comes from the parallel computing ability of the generation module and the constraint pre-screening mechanism of the optimization algorithm, effectively avoiding the invalid path dependence in manual plan exploration (as shown in Figure 1).\u003c/p\u003e\n\u003cp\u003eAt the aesthetic evaluation level, a composite evaluation method that combines double-blind expert review and user perception is adopted. The aesthetic consistency score of the model output scheme reaches 82.5%, which is a 51.9% increase compared to 54.3% of the traditional method. This leap benefits from the dynamic alignment ability of the cross-modal attention mechanism to cultural context and functional requirements (as shown in Table 1).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 1.\u0026nbsp;\u003c/strong\u003eComparison of aesthetic evaluation consistency (%)\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"570\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003eEvaluation dimension\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003eTraditional solution\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eHuman-machine collaboration solution\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e\u0026Delta; value\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003eCohen\u0026rsquo;s d\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003ep value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003eEmotional resonance degree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e3.7\u0026plusmn;0.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e4.5\u0026plusmn;0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e+0.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e1.05\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e\u0026lt;0.001***\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003eFunctional convenience\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e3.4\u0026plusmn;1.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e4.2\u0026plusmn;0.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e+0.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.87\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.002**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003eVisual appeal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e3.8\u0026plusmn;0.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e4.4\u0026plusmn;0.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e+0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.90\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.013*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003eSpatial readability\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e3.5\u0026plusmn;1.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e4.1\u0026plusmn;0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e+0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.73\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e0.021*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003eCase empirical research\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe application verification of the human-machine collaborative model in typical scenarios reveals the differentiated value paths of intelligent technologies enabling spatial aesthetics. In the case of cultural exhibition hall design, the generation module outputs 8 stylized structural schemes such as Neo-Futurism and Deconstructivism through style interpolation technology. Among them, the theme of \u0026ldquo;broken line metaphor\u0026rdquo;was selected by the designer because it fits the industrial heritage context of the site. The system then initiates multi-physical field coupling optimization: based on the flow line analysis of pedestrian flow simulation data, the width of the main passage is dynamically adjusted from the initial plan of 3.2 meters to an elastic range of 2.8 - 4.5 meters, enhancing the fun of spatial exploration while ensuring evacuation efficiency; for the lighting parameter optimization, a non-uniform B-spline curve is introduced to control the light attenuation gradient, keeping the illuminance of the exhibition wall within the optimal viewing range of 150 - 200 lux, and the power density is reduced by 22% compared to traditional designs. This dynamic tuning mechanism liberates designers from technical details and allows them to focus on the construction of spatial narratives (as shown in Figure 2).\u003c/p\u003e\n\u003cp\u003eCompared with the morphological innovation challenges of cultural exhibition halls, the case of the atrium renovation of commercial spaces highlights the value transformation ability driven by user needs. By analyzing 2,350 sets of user eye movement data and heart rate variability records, the algorithm identified a preference cluster for \u0026ldquo;natural light and shadow\u0026rdquo;, and based on this, an optimized plan with a parametric wooden grille dome was generated. The size of the grille units changes dynamically according to the solar trajectory (the porosity is 15% at noon in summer \u0026rarr; 45% in the morning in winter). While controlling the glare index, it allows the spatial illuminance to fluctuate naturally within the comfortable range of 300 - 500 lux. During the construction phase, the robot prefabrication and assembly technology was adopted to control the error of 7,200 special-shaped components within \u0026plusmn;1.5 mm. Finally, the project cost was reduced by 18% (as shown in Table 2), and the average user stay time increased by 24 minutes.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2.\u0026nbsp;\u003c/strong\u003eComparison of the benefits of commercial space renovation\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 159px;\"\u003e\n \u003cp\u003eindicator\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eTraditional solution\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 196px;\"\u003e\n \u003cp\u003eHuman-machine collaboration solution\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 121px;\"\u003e\n \u003cp\u003eRate of change\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 159px;\"\u003e\n \u003cp\u003eConstruction cost (in ten thousand yuan)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e850\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 196px;\"\u003e\n \u003cp\u003e697\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 121px;\"\u003e\n \u003cp\u003e-18.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 159px;\"\u003e\n \u003cp\u003eAverage daily passenger flow (person-times)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e2300\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 196px;\"\u003e\n \u003cp\u003e2890\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 121px;\"\u003e\n \u003cp\u003e+25.7%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 159px;\"\u003e\n \u003cp\u003eUser satisfaction (on a 5-point scale)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e3.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 196px;\"\u003e\n \u003cp\u003e4.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 121px;\"\u003e\n \u003cp\u003e+19.4%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 159px;\"\u003e\n \u003cp\u003eEnergy consumption intensity（kWh/m\u0026sup2;）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e115\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 196px;\"\u003e\n \u003cp\u003e93\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 121px;\"\u003e\n \u003cp\u003e-19.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003eComparison between subjective and objective evaluations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe validation of the effectiveness of the human-machine collaborative model requires simultaneously examining the objective accuracy of algorithm predictions and the matching degree of users\u0026rsquo; subjective perception. The correlation analysis between expert scores and algorithm predicted values shows that there is a significant positive correlation between the two in the evaluation dimension of spatial aesthetic quality (Pearson\u0026rsquo;s r = 0.79, p \u0026lt; 0.01), indicating that the intelligent system can effectively capture the core aesthetic elements in professional evaluations. The distribution trend of the scatter plot reveals that when the algorithm predicted value is higher than 80 points (as shown in Figure 3), the standard deviation of expert scores narrows to \u0026plusmn;3.2 points, indicating that the model has higher judgment stability in the identification of high-quality solutions.\u003c/p\u003e\n\u003cp\u003eThe results of the user satisfaction survey further confirm the practical value of the model. The human-machine collaboration solution is significantly superior to the traditional design method in sub-dimensions such as emotional resonance (4.5 vs. 3.7) and functional convenience (4.2 vs. 3.4) (independent samples t-test, p\u0026lt;0.001). Moreover, the standard deviation of overall satisfaction decreased from 0.82 in the control group to 0.51, reflecting that intelligent technology effectively reduces the evaluation divergence caused by individual aesthetic differences (as shown in Table 3).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 3.\u0026nbsp;\u003c/strong\u003eComparison of user satisfaction by dimension (on a 5-point scale)\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 106px;\"\u003e\n \u003cp\u003eEvaluation dimension\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003eTraditional solution\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003eHuman-machine collaboration solution\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e\u0026Delta; value\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 112px;\"\u003e\n \u003cp\u003ep value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 106px;\"\u003e\n \u003cp\u003eEmotional resonance degree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e3.7\u0026plusmn;0.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003e4.5\u0026plusmn;0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e+0.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 112px;\"\u003e\n \u003cp\u003e\u0026lt;0.001***\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 106px;\"\u003e\n \u003cp\u003eFunctional convenience\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e3.4\u0026plusmn;1.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003e4.2\u0026plusmn;0.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e+0.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 112px;\"\u003e\n \u003cp\u003e0.002**\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 106px;\"\u003e\n \u003cp\u003eVisual appeal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e3.8\u0026plusmn;0.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003e4.4\u0026plusmn;0.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e+0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 112px;\"\u003e\n \u003cp\u003e0.013*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 106px;\"\u003e\n \u003cp\u003eSpatial readability\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e3.5\u0026plusmn;1.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 177px;\"\u003e\n \u003cp\u003e4.1\u0026plusmn;0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 87px;\"\u003e\n \u003cp\u003e+0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 112px;\"\u003e\n \u003cp\u003e0.021*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e"},{"header":"Discussion","content":"\u003cdiv id=\"Sec21\" class=\"Section2\"\u003e\u003ch2\u003eModel effectiveness\u003c/h2\u003e\u003cp\u003eThe technical effectiveness of the human-machine collaboration model is reflected in its ability to bidirectionally expand the design cognitive framework. By deconstructing the implicit knowledge graph in historical cases, the intelligent system can establish a topological association network of cross-temporal and cross-cultural symbols. For example, the mechanical structure of the dougong in traditional architecture is deconstructed into a node connection rule base and then reorganized into an overhanging form that conforms to modern mechanical principles. This generation mechanism of non-explicit associations breaks through the time scale of designers\u0026rsquo; individual experience and the spatial and cultural boundaries. In the digital reconstruction case of the New Museum of Suzhou, the algorithm-generated variant schemes of the eaves tile patterns through potential space exploration not only maintain the morphological rhythm of the Coiled chi-dragon patterns of the Warring States Period from the Warring States period but also adapt to the assembly logic of contemporary glass curtain walls through topological optimization. Such schemes were initially judged by designers as unconventional options because they exceeded the scope of traditional experience. However, their cultural translation effectiveness was confirmed after parametric verification.\u003c/p\u003e\u003cp\u003eMeanwhile, the model ensures the decision-making dominance of humanistic values through a dynamic permission allocation mechanism. During the scheme deepening stage, the system automatically marks the culturally sensitive elements in the schemes generated by the algorithm (such as the spatial proportion of religious symbols, the deviation degree of historical authenticity, etc.), triggering the manual review process. Taking the renewal design of the Xi\u0026rsquo;an Small Wild Goose Pagoda area as an example, although the digital simplification scheme of the Tang-style roof ridge proposed by the algorithm passed the structural rationality verification, it was vetoed by the designer based on the data of historical literature research and the parameter rollback was initiated because the upturned angle of the eaves deviated from the threshold of the characteristics of the Tang Dynasty architectural style. This collaborative paradigm of \u0026ldquo;machine inspiration - human judgment\u0026rdquo; not only avoids the risk of style convergence caused by algorithmic hegemony, but also provides empirical support beyond the individual knowledge reserve for the designer\u0026rsquo;s critical decision-making through the cross-cultural knowledge retrieval function of the intelligent system, ultimately achieving the dialectical unity of technological innovation and cultural inheritance.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec22\" class=\"Section2\"\u003e\u003ch2\u003eComparison with traditional methods\u003c/h2\u003e\u003cp\u003eThe core breakthrough of the human-machine collaborative model compared with traditional environmental art design methods lies in its dynamic response ability to complex constraint systems and the depth of demand mining. In the traditional process, designers often handle hard constraints such as energy conservation and accessibility through linear superposition, resulting in multi-objective optimization falling into the local optimum trap. For example, in the aging-friendly renovation of old communities, when manually balancing the lighting requirements and thermal performance, conventional parametric tools can only generate a limited number of approximate solutions. In contrast, this model introduces implicit constraint modeling technology, encoding regulatory provisions and regional climate data into an adaptive weight matrix, enabling the algorithm to simultaneously optimize 23 parameters such as the angle of sunshading components and the turning radius of wheelchairs during the plan generation stage, and achieving the automatic approximation of the Pareto front.\u003c/p\u003e\u003cp\u003eA more significant advantage lies in the ability to capture long-tail demands: Through the sentiment analysis of user behavior data and social media comments, the model can identify the needs of special groups that account for less than 5% (such as the spatial safety preferences of autistic children), and generate customized design elements such as high window sills and soft partitions accordingly. This is difficult to achieve in the traditional expert experience-driven model. At the same time, through the dynamic weight allocation mechanism, this model can real-time adjust the optimization priorities of function, aesthetics, and cost during the scheme deepening stage (such as automatically increasing the weight of modular assembly during the construction stage), enabling the identification and resolution of design conflicts to be advanced to the virtual verification link. As a result, the change rate of the construction plan is reduced from the industry average of 32\u0026ndash;14.5%. This data confirms the role of dynamic weight allocation in integrating the design-construction value chain - the algorithm not only generates solutions but also anticipates implementation risks through the flexible adaptation of constraints, significantly reducing the marginal cost of project iteration.\u003c/p\u003e\u003cp\u003eHowever, the model still faces structural challenges in terms of generation quality when dealing with small-sample scenarios. When the scarcity of training data and the complexity of design tasks create dual pressures (such as ultra-low-frequency projects like the renovation of Qiang watchtowers or burn rehabilitation centers), the algorithm may produce abnormal solutions with cultural context misalignment or functional logic conflicts. For example, in the case of the renovation of a southwestern ethnic minority village, due to the lack of samples of stilt building structures in the system, the generated elevated floor structure met modern seismic resistance specifications, but misapplied Dai architectural symbols to a Miao settlement. This reveals the shallow association defect of existing technologies with domain knowledge. There is an urgent need to construct a knowledge graph system that integrates unstructured knowledge such as construction methods and craftsmen\u0026rsquo;s formulas. In the future, through domain knowledge distillation technology, expert experience can be transformed into a tunable rule loss function, enabling the algorithm to maintain cultural adaptability even with a small number of samples, thus truly achieving the deep integration of intelligent design and humanistic values.\u003c/p\u003e\u003cdiv id=\"Sec23\" class=\"Section3\"\u003e\u003ch2\u003eReflection on technological ethics\u003c/h2\u003e\u003cp\u003eThe ethical risks of intelligent design technology coexist with its innovative potential. The problem of algorithmic bias manifests as a crisis of implicit style convergence in the field of spatial aesthetics. When the training dataset overly relies on cases of International Style modernist architecture, the generative system may generalize the visual principle of \u0026ldquo;less is more\u0026rdquo; into a universal aesthetic standard, resulting in regional cultural symbols being reduced to superficial decorative elements. For example, a clustering analysis of 48 intelligent generation plans for traditional village renewal found that 76% of the plans adopted a combination of standardized pitched roofs and glass curtain walls, while the utilization rate of local materials was less than 12%. This implicit bias essentially constitutes a technological and cultural hegemony in the digital age. Therefore, it is necessary to embed a diversity guarantee mechanism in the algorithm architecture, including establishing a cultural sensitivity detection module - identifying the density distribution of style features in the plan through a semantic network. When the probability of a certain style exceeds the preset threshold, the potential space perturbation algorithm is automatically triggered to force the exploration of the non-dominated solution region, thereby maintaining the diversity balance of the creative ecosystem.\u003c/p\u003e\u003cp\u003eThe deeper impact of technology ethics is reflected in the role reconstruction of the design subject. When the intelligent system undertakes more than 60% of the tasks of form generation and performance optimization, the core ability of designers is shifting from spatial form shaping to value strategy formulation. In the design of a certain science and technology innovation center in Xiongan New Area, the team is no longer confined to single form deliberation. Instead, by adjusting the algorithm weight matrix, it dynamically balances the multi-dimensional game relationships among the cost constraints of investors, the collaboration needs of scientific researchers, and the public\u0026rsquo;s expectations for open spaces.\u003c/p\u003e\u003cp\u003eThis transformation requires a paradigm-level reconstruction of professional ability standards: designers need to master new skills such as human-computer interaction protocol design, algorithm ethics review, and interdisciplinary knowledge distillation. Traditional aesthetic literacy and spatial operation ability are transformed into the ability to optimize the prompt engineering of intelligent systems and culturally decode the output results. The education system urgently needs to establish a cultivation framework for \u0026ldquo;critical technology thinking\u0026rdquo; so that designers can identify the value load in the algorithm black box and adhere to the leading position of humanistic spirit in human-machine collaboration.\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThrough theoretical construction and practical verification, this research reveals the innovative path and transformation boundaries of intelligent technology enabling environmental art design. At the theoretical level, the proposed \u0026ldquo;aesthetics-technology-human factors\u0026rdquo; ternary collaborative framework breaks through the binary opposition mindset of form and function in traditional design theories. By establishing a cross-modal attention mechanism and a dynamic weight allocation algorithm, it realizes the real-time interactive mapping of the cultural semantic network, engineering parameter system, and user behavior map, providing a computable cognitive interface for the digital translation of spatial aesthetics. This theoretical breakthrough not only explains the internal mechanism of creative emergence in the process of human-computer collaboration but also promotes the transformation of the environmental art discipline from an experience-dependent paradigm to a data-knowledge dual-driven paradigm by constructing a quantitative channel for design value transfer.\u003c/p\u003e\u003cp\u003eIn the practical dimension, model validation shows that this framework has significant technology spillover effects. Taking the renewal of smart city blocks as an example, by embedding real-time traffic flow and microclimate data, the system can dynamically adjust the height-width ratio and greening penetration rate of the street interface. While maintaining the integrity of the historical features, it can reduce the heat island intensity by 1.2\u0026deg;C. In the scenario of digital restoration of Dunhuang mural caves, the algorithm, through feature extraction of multi-spectral images and topological reconstruction of missing parts, increases the restoration efficiency of traditional craftsmen by 4 times, and the color restoration accuracy reaches 93%. These cross-field application cases confirm the tool value of the model in balancing cultural inheritance and technological innovation, providing an expandable technological foundation for new urbanization and the activation of cultural heritage.\u003c/p\u003e\u003cp\u003eIt is worth noting that the adaptability of the current technical system in cross-cultural contexts still has limitations. When dealing with aesthetic systems dominated by irregular forms (such as African tribal art or Austronesian architecture), algorithms are vulnerable to the cultural biases of training data, posing a risk of symbol misplacement. Future research needs to focus on breaking through the bottleneck of domain knowledge transfer under small-sample conditions. By constructing an interpretable style feature decoupling model, cross-cultural adaptation of aesthetic elements can be achieved. At the same time, develop a low-code interactive platform for designers, encapsulate underlying functions such as latent space exploration and parameter sensitivity analysis into visual modules, enabling intelligent design technology to break through the professional barriers of computer scientists and truly become a democratized tool for environmental art innovation.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets generated and/or analysed during the current study are not publicly available due to the proprietary nature of the intelligent design model and the core algorithms but are available from the corresponding author on reasonable request（[email protected]）.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eShaoqing Wang:\u0026nbsp;Conceptualization, Methodology, Software, Formal Analysis, Writing \u0026ndash; Original Draft Preparation.\u003cbr\u003e\u0026nbsp;Jian Yao: Methodology, Validation, Investigation, Data Curation, Writing \u0026ndash; Review \u0026amp; Editing.\u003c/p\u003e\n\u003cp\u003eShaoqing Wang proposed the research concept, developed the human-machine symbiotic intelligent design model, designed the generative adversarial network for spatial semantic analysis, and drafted the manuscript. Jian Yao contributed to the construction of the multi-objective dynamic optimization framework, implemented the augmented reality collaborative interface, conducted empirical validation, and performed data analysis. Both authors participated in the interpretation of results, manuscript revision, and approval of the final version.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors received no funding for this work.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics statement\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study did not involve human participants requiring ethical approval beyond standard informed consent for data collection. All participant data were collected anonymously and used solely for academic research purposes. No personally identifiable information was retained. The research procedures were conducted in accordance with the ethical standards of Krirk University and the principles outlined in the Declaration of Helsinki. All authors confirm that there are no ethical conflicts related to this work.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eCai Qing, Liao Wenjie, Xue Hongjing, Huang Shengnan, Lu Xinzheng \u0026amp; Zhang Shuo.(2024). Application Method and Case Study of the Generative Intelligent Design Tool for Shear Wall Structure Schemes. Building Structure, 1 - 9. DOI: 10.19701/j.jzjg.LS243053.\u003c/li\u003e\n \u003cli\u003eCao Rongrong, Liu Lin, Yu Yandong \u0026amp; Wang Hailong.(2024). A review of research on large language models integrating knowledge graphs. Application Research of Computers, 1 - 14. DOI: 10.19734/j.issn.1001 - 3695.2024.12.0532.\u003c/li\u003e\n \u003cli\u003eLi Hui, Wang Junyin, Cheng Yuanzhi, Liu Jian, Zhao Guowei \u0026amp; Chen Shuangmin. (2024). A 3D object detection method based on image semantic feature guidance and cross - modal fusion of point clouds. Journal of Computer - Aided Design \u0026amp; Computer Graphics (05), 734 - 749.\u003c/li\u003e\n \u003cli\u003eLi Yubin. (2025). The cultural style, spatial narrative and image expression of documentaries on the Yellow River theme under the theory of cultural semiotics. Movie Review (05), 55 - 60. DOI: 10.16583/j.cnki.52 - 1014/j.2025.05.006.\u003c/li\u003e\n \u003cli\u003eLiang Yinjie, Nan Xinyuan, Cai Xin, Li Yunpeng \u0026amp; Gou Haiguang.(2024). Bridge Defect Detection Based on Data Augmentation and Improved YOLOv8. Journal of Guangxi Normal University (Natural Science Edition), 1 - 16. DOI: 10.16088/j.issn.1001 - 6600.2024071003.\u003c/li\u003e\n \u003cli\u003eMa Yongjie \u0026amp; Chen Min. (2022). Dynamic Multi - objective Optimization Algorithm Based on Kalman Filter Prediction Strategy. Journal of Jilin University (Engineering and Technology Edition) (06), 1442 - 1458. DOI: 10.13229/j.cnki.jdxbgxb20210082.\u003c/li\u003e\n \u003cli\u003eNie Ying. (2024). Research on the Environmental Art Design Industry and Education in the AI Era. Art Design Research (05), 120 - 128.\u003c/li\u003e\n \u003cli\u003ePei Xuan. (2023). The Significance Structure Expression of \u0026ldquo;Sense of Place\u0026rdquo; and Spatial Aesthetics. Social Sciences in Yunnan (01), 153 - 163.\u003c/li\u003e\n \u003cli\u003ePeng Lan. (2024). The Drift of Boundaries and the Guardianship of Subjectivity: Reflection on the Issue of Human Subjectivity under the Trend of Intelligent Content Production. Journal of Soochow University (Philosophy and Social Science Edition) (05), 159 - 170. DOI: 10.19563/j.cnki.sdzs.2024.05.015.\u003c/li\u003e\n \u003cli\u003eQin Junhui. (2020). Research on the environmental art design system based on virtual reality technology. Modern Electronics Technique (13), 92 - 94 + 98. DOI: 10.16652/j.issn.1004 - 373x.2020.13.022.\u003c/li\u003e\n \u003cli\u003eTang Yi. (2025). The Mechanism and Path of the Inheritance and Development of Excellent Traditional Chinese Culture from the Perspective of Design Aesthetics. Jiangsu Social Sciences (01), 202 - 210 + 244. DOI: 10.13858/j.cnki.cn32 - 1312/c.20250121.003.\u003c/li\u003e\n \u003cli\u003eWang Shishun. (2023). Research on the Creative Potential of Generative Adversarial Networks in the Field of Art Design. Packaging Engineering (S1), 24 - 28 + 41. DOI: 10.19554/j.cnki.1001 - 3563.2023.S1.004.\u003c/li\u003e\n \u003cli\u003eWang Wenzhong, Zhang Shusheng, Yu Suihuai \u0026amp; Chen Dengkai. (2021). Optimization of the cabin layout of submersibles based on deconstruction and reconstruction thinking. Computer Integrated Manufacturing Systems (07), 2045 - 2052. DOI: 10.13196/j.cims.2021.07.019.\u003c/li\u003e\n \u003cli\u003eWang Xingcheng, Wang Haoyu, Pan Xinyu, Tao Fang, Zheng Xingxing \u0026amp; Cao Shuang.(2025). Applications of Machine Learning in the Field of Environmental Microplastics: Classification and Identification, Quantitative Analysis, and Adsorption Performance Prediction. China Environmental Science, 1 - 15. DOI: 10.19674/j.cnki.issn1000 - 6923.20250507.002.\u003c/li\u003e\n \u003cli\u003eXiong Liang. (2024). Concerns, Optimization and Future Trends of the Educational Application of Affective Computing in the Era of Intelligent Technology. Open Education Research (06), 66 - 71. DOI: 10.13966/j.cnki.kfjyyj.2024.06.008.\u003c/li\u003e\n \u003cli\u003eXu Ming. (2025). The digital logic and its implementation path for improving the effectiveness of grass - roots governance. Inner Mongolia Social Sciences (03), 55 - 61. DOI: 10.14137/j.cnki.issn1003 - 5281.2025.03.008.\u003c/li\u003e\n \u003cli\u003eXu Muchuan \u0026amp; Wu Zijun. (2020). Research on the iteration and design of complex commercial complexes. Art Journal (02), 102 - 107.\u003c/li\u003e\n \u003cli\u003eXu Qianyao, Wu Qiong \u0026amp; Gong Wei. (2022). Analysis and Inspiration of Intelligent Design Methods Based on Aesthetic Computation. Decoration (09), 17 - 22. DOI: 10.16272/j.cnki.cn11 - 1392/j.2022.09.031.\u003c/li\u003e\n \u003cli\u003eZhang Delu.(2025). A Study on the Applicability of Social Semiotics in the Construction of Multimodal Discourse. Journal of Beijing International Studies University, 1 - 13.\u003c/li\u003e\n \u003cli\u003eZhang Jingyu, Liu Nianxiong, Wang Shanshan \u0026amp; Yan Shurui. (2022). Energy - saving Performance - oriented Intelligent Generation Design Method and Tool Platform for Residential Buildings. Architectural Journal (02), 22 - 27. DOI: 10.19819/j.cnki.ISSN0529 - 1399.202202004.\u003c/li\u003e\n \u003cli\u003eZhang Kaixuan, Cai Guoyong \u0026amp; Zhu Kunri. (2023). Visual Perception Recommendation System with Enhanced Image Aesthetic Information. Computer Science (S2), 285 - 292.\u003c/li\u003e\n \u003cli\u003eZhang Wenting, Liu Mandan, Zhu Baoxu \u0026amp; Xie Lizhi.(2024). Generation of 3D Facial Expression Animation Based on Interactive Evolutionary Algorithm. Journal of East China University of Science and Technology (Natural Science Edition), 1 - 15. DOI: 10.14135/j.cnki.1006 - 3080.20240831001.\u003c/li\u003e\n \u003cli\u003eZhou Xiang, Cui Lichen, Chen Suqing \u0026amp; Tan Zimei. (2024). Visual Salience Analysis of Imagery Elements in Historical Blocks Based on Eye - tracking: Taking Yongqingfang in Guangzhou as an Example. Modern Urban Research (10), 101 - 107.\u003c/li\u003e\n \u003cli\u003eZuo Wenjin, He Xiaogang, Wen Chuanzhen \u0026amp; Liu Lijun. (2022). Research on the construction of the evaluation index system for the quality of big data resources - Analysis of library big data from the perspective of user perception. Price:Theory \u0026amp; Practice (08), 55 - 58. DOI: 10.19851/j.cnki.CN11 - 1010/F.2022.08.445.\u003c/li\u003e\n\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":"Intelligent design, Spatial aesthetics, Human-machine Collaboration, Generative adversarial network, Multi-objective optimization, Multimodal interaction, Aesthetic computing","lastPublishedDoi":"10.21203/rs.3.rs-7464844/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7464844/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eDriven by intelligent technologies, environmental art design is undergoing a paradigm reconstruction from experience dependence to a data-knowledge dual-engine support model. Aiming at the structural defects of the traditional paradigm, such as efficiency bottlenecks, rigid creative boundaries, and sluggish dynamic responses, this study constructs an intelligent design model based on the logic of human-machine symbiosis. Through the integration of three technical modules: the spatial semantic analysis of the generative adversarial network, the multi-objective dynamic optimization of the parametric system, and the immersive collaboration of the augmented reality interface, a closed-loop innovation chain of aesthetic feature vectorization expression - algorithm generation - manual adjustment is formed. Empirical research shows that in typical scenarios such as the activation of cultural heritage and intelligent complexes, this model can increase the efficiency of creative plan generation by 42%, achieve an 82.5% consistency in the evaluation of interdisciplinary experts, enhance the intensity of user emotional resonance by 60% compared with traditional methods, and reduce the construction plan change rate by 55%. This technical framework not only reveals the computable interaction mechanism between algorithm logic and aesthetic laws, but also provides a transformation path with both cognitive breakthroughs and engineering implementation potential for the environmental art field by establishing dynamic weight allocation and ethical review rules. Its methodological system can be extended to the sustainable construction practices of complex systems such as urban renewal and digital cultural heritage protection.\u003c/p\u003e","manuscriptTitle":"Intelligent design empowers spatial aesthetics: construction of a man-machine collaborative innovation model for environmental art and case verification","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-22 10:18:56","doi":"10.21203/rs.3.rs-7464844/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":"2df5bc49-12e0-427e-b1cb-886f3139ae7e","owner":[],"postedDate":"September 22nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":54600932,"name":"Physical sciences/Engineering"},{"id":54600933,"name":"Physical sciences/Mathematics and computing"}],"tags":[],"updatedAt":"2026-03-11T14:12:06+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-22 10:18:56","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7464844","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7464844","identity":"rs-7464844","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}