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.