In this study, an improved deep learning model is proposed to explore the complex interactions between the road environment and driver’s behaviour throughout the generation of a graphical representation. The proposed model consists of an unsupervised Denoising Stacked Autoencoder (SDAE) able to provide output layers in RGB colors. The dataset comes from an experimental driving test where kinematic measures were tracked with an in-vehicle GPS device. The graphical outcomes reveal the method ability to efficiently detect patterns of simple driving behaviors, as well as the road environment complexity and some events encountered along the path.

The emergence of Multimodal Large Language Models (MLLMs) has made methods of artificial intelligence accessible to the general public in a conversational way. It offers tools for the automated visual assessment of the quality of a built environment for professionals of urban planning without requiring specific technical knowledge on computing. We investigated the capability of MLLMs to perceive urban environments based on images and textual prompts. We compared the outputs of several popular models—ChatGPT, Gemini and Grok—to the visual assessment of experts in Architecture, Engineering and Construction (AEC) in the context of a real estate construction project. Our analysis was based on subjective attributes proposed to characterize various aspects of a built environment. Four urban identities served as case studies, set in a virtual environment designed using professional 3D models. We found that there can be an alignment between human and AI evaluation on some aspects such as space and scale and architectural style, and more general accordance in environments with vegetation. However, there were noticeable differences in response patterns between the AIs and AEC experts, particularly concerning subjective aspects such as the general emotional resonance of specific urban identities. It raises questions regarding the hallucinations of generative AI where the AI invents information and behaves creatively but its outputs are not accurate.

Identifying the spatio-temporal patterns of people activities in urban areas is key to effective urban planning; it can be used in real-estate projects to predict their future impacts on behavior in surrounding accessible areas. LaVallée is a large construction project recently started in Paris’s suburb; it is a new district due in 2024. The paper is in the field of urban planning, aiming at developing a method making it possible to model the potential visits of the various equipment and public spaces of the district, by mobilizing data from census at the departmental level, and the layout of shops and activities as defined by the real-estate project. This model takes into account the flow of external visitors, estimated realistically based on the pre-project movements in the areas of influence of LaVallée. In this paper, we propose an activity-based model methodology to determine trips and their purpose at a mesoscopic scale including the city and surrounding areas, in the current baseline scenario. This travel demand is required to estimate potential external visitors of the future district. A first demonstration shows that the model correctly represents the current demands and allows the forecast of future demand in the area.

Human impression plays a crucial role in effectively designing infrastructures that support active mobility such as walking and cycling. By involving users early in the design process, valuable insights can be gathered before physical environments are constructed. This proactive approach enhances the attractiveness and safety of designed spaces for users. This study conducts an experiment comparing real street observations with immersive virtual reality (VR) visits to evaluate user perceptions and assess the quality of public spaces. For this experiment, a high-resolution 3D city model of a large-scale neighborhood was created, utilizing Building Information Modeling (BIM) and Geographic Information System (GIS) data. The model incorporated dynamic elements representing various urban environments: a public area with a tramway station, a commercial street with a road, and a residential playground with green spaces. Participants were presented with identical views of existing urban scenes, both in reality and through reconstructed 3D scenes using a Head-Mounted Display (HMD). They were asked questions related to the quality of the streetscape, its walkability, and cyclability. From the questionnaire, algorithms for assessing public spaces were computed, namely Sustainable Mobility Indicators (SUMI) and Pedestrian Level of Service (PLOS). The study quantifies the relevance of these indicators in a VR setup and correlates them with critical factors influencing the experience of using and spending time on a street. This research contributes to understanding the suitability of these algorithms in a VR environment for predicting the quality of future spaces before occupancy.

With the rapid growth in social network comments, the need for more effective methods to classify their polarity—negative, neutral, or positive—has become essential. Sentiment analysis, powered by natural language processing, has evolved significantly with the adoption of advanced deep learning techniques. Long Short-Term Memory networks capture long-range dependencies in text, while transformers, with their attention mechanisms, excel at preserving contextual meaning and handling high-dimensional, semantically complex data. This study compares the performance of sentiment analysis models based on LSTM and BERT architectures using key evaluation metrics. The dataset consists of business reviews from the Yelp Open Dataset. We tested LSTM-based methods against BERT and its variants—RoBERTa, BERTweet, and DistilBERT—leveraging popular pipelines from the Hugging Face Hub. A class-by-class performance analysis is presented, revealing that more complex BERT-based models do not always guarantee superior results in the classification of Yelp reviews. Additionally, the use of bidirectionality in LSTMs does not necessarily lead to better performance. However, across a diversity of test sets, transformer models outperform traditional RNN-based models, as their generalization capability is greater than that of a simple LSTM model.