In 2024, 1,040 accidents were recorded on Spanish roads, in addition to minor collisions and other driving problems. The causes of these accidents include speeding, adverse weather conditions and substance abuse, but also distraction and stressful situations that can be mitigated by improving infrastructure design, technologies to assist drivers and road safety policies.
A study in which the Universitat Oberta de Catalunya (UOC) took part analyzes how visual elements influence drivers’ stress levels, and identifies factors that negatively affect the driving experience. Its findings pave the way for the development of smart driving assistants and the planning of city streets with fewer stress triggers.
Published in IEEE Transactions on Affective Computing, the paper “Analyzing the Visual Road Scene for Driver Stress Estimation” presents the conclusions of a research project led by Cristina Bustos, a researcher of the Artificial Intelligence for Human Well-being (AIWELL) group, which is affiliated to the UOC’s research unit on Digital Health, Health and Well-being.
The research also involved faculty members from the UOC’s Faculty of Computer Science, Multimedia and Telecommunications, Àgata Lapedriza, AIWELL’s lead researcher, and Albert Solé, a CoSIN3 researcher. Javier Borge, leader of the Complex Systems group (CoSIN3) also contributed to the project, together with Neska Elhaouij and Rosalind Picard, researchers from the Media Lab at the Massachusetts Institute of Technology (MIT).
The importance of the road landscape as a stress factor
The things we see on roads and the areas around them are a key factor in traffic accidents and greatly influence the well-being and health of those who use their cars daily. So, addressing the causes of stress among drivers has been central to a number of studies in recent years.
The research carried out by the UOC analyzes these factors, taking only visual data into account, ignoring physiological signals, facial analysis or recordings of vehicle maneuvers. This is the first time that a study of this nature has focused solely on the visual aspect.
“Up to now, we haven’t taken into account that we drive in a visual context and that conditions in the urban setting matter as they affect the driver’s stress level. Our study is the first to analyze the visual context of urban scenery as an additional source of data to estimate stress,” said the expert.
The UOC team used an AI model that simultaneously evaluates traffic conditions, the presence of pedestrians and features of the urban environment in real-world settings to conduct a large-scale study of the visual landscape.
Several machine learning models with different levels of complexity were used. They included support vector machines (SVMs) and convolutional neural networks (CNNs), which analyzed individual images, and temporal segment networks (TSNs), which evaluated videos.
“Our approach studies the context of the road, analyzes how the driving environment affects driver stress and helps predict it. We have empirically demonstrated that analyzing the visual environment provides valuable contextual information about the road environment, such as traffic density, the urban landscape and the presence of pedestrians,” Bustos said.
“This information complements other data sources and is essential for a better understanding of the factors that influence stress levels and how urban design can impact road safety. Our study demonstrates, for the first time, that context is a significant source of information that can be processed,” she added.
Pedestrians, other vehicles and road signs generate stress
The UOC team concluded that the visual context of the road plays a fundamental role in causing driver stress and was able to determine which specific elements most influence the driving experience.
The analysis of the AI model revealed that the presence of pedestrians and moving vehicles (especially larger ones such as lorries) are among the factors that generate most stress. Added to this are urban elements that can distract drivers, such as signs, advertising posters and pedestrian crossings.
“All of these elements significantly influence the high stress levels of the drivers studied, by increasing the complexity of the experience and their cognitive load,” Bustos explained.
Potential practical applications
These findings can serve as a guide for the design of urban infrastructure and policies aimed at reducing stress-inducing factors. For example, they could be the basis for improvements in signage, traffic management systems in congested areas, or the design of safer intersections.
“By identifying which elements are the most stressful, urban planners and traffic authorities can take measures to mitigate these effects, contributing to greater road safety,” said Bustos.
There is also the possibility of developing driver assistance systems that can monitor the environment in real time and alert the driver or activate safety mechanisms when potentially stressful conditions are detected.
“At the moment, there are no immediate plans for the practical application of the study, as it was conducted with a limited number of drivers. However, the results provide a promising basis for continuing research in this area and exploring its application to driver assistance systems,” Bustos said.
The next steps in this line of research are to expand and diversify the data, explore multimodal models that incorporate other types of non-invasive data (such as information on the vehicle), and refine AI interpretation techniques to better understand the mechanisms underlying stress.
More information:
Cristina Bustos et al, Analyzing the Visual Road Scene for Driver Stress Estimation, IEEE Transactions on Affective Computing (2025). DOI: 10.1109/TAFFC.2025.3539003
Open University of Catalonia
Citation:
AI model pinpoints sources of driver stress, paving the way for smart driving assistants (2025, May 26)
retrieved 26 May 2025
from https://techxplore.com/news/2025-05-ai-sources-driver-stress-paving.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.
