Improving urban cyclist safety and skills: Integrating a multiagent system and virtual reality training simulations

Abstract

This study investigates how integrating immersive virtual reality with a multi-agent system can improve urban cyclist training by adapting learning experiences to individual performance. Addressing the challenge of preparing cyclists for complex urban environments, the research explores whether an adaptive VR-based system can enhance hazard perception, decision-making, and compliance with traffic rules. The proposed system leverages a context-aware multi-agent framework that dynamically adjusts traffic density, environmental conditions, and scenario complexity based on user behaviour. This personalized approach ensures that training remains challenging yet accessible, fostering progressive skill acquisition in a safe, controlled simulation environment. A preliminary evaluation was conducted with eight participants over a month-long training period. Results indicated improvements in reaction times, safety distance compliance, and overall traffic rule adherence. The system’s adaptability and ability to integrate into existing urban training programs suggest its potential as a scalable, data-driven tool for cyclist education.