BiciVR: A Software Engineering Framework for AI-Driven Bicycle Mobility Risk Simulation

Abstract

The rise of micromobility in the Guadalajara Metropolitan Area (GMA) introduces new road safety challenges. This study proposes a software engineering framework that integrates virtual reality and data science to identify cyclist behavior and accident risk scenarios. A gamified simulation in Unity 3D, using Oculus Meta Quest headsets, allowed 36 participants to navigate realistic traffic conditions. Data on collisions and violations were analyzed using machine learning and heatmaps. Results show that 4.31% of violations led to collisions, with intersections and traffic density as key risk factors. This framework supports targeted interventions for safer urban mobility.