AI-Driven Virtual Coach for Immersive Adaptive VR Cycling Training in Young Students with Disabilities

This paper presents an AI-driven Virtual Coach system for inclusive adaptive cycling training in young adults with disabilities (18–34 years old). The platform combines a physical stationary bicycle, a Unity-based Virtual Reality (VR) cycling environment, and real-time markerless motion capture using The Captury Live. Full-body kinematic data are streamed to a digital avatar, while virtual terrain changes are synchronized with dynamic bicycle resistance to simulate uphill and downhill riding in a safe and controlled setting. The AI-driven coach monitors cadence, trunk alignment, pedaling symmetry, and fatigue-related indicators, providing simplified visual and audio feedback adapted to users’ cognitive and motor abilities. The system was evaluated in a university living-lab context with 66 participants from local associations, supported by educators. Usability, tolerability, and functional changes were assessed through the System Usability Scale (SUS), an adapted cybersickness questionnaire based on the SSQ, and the International Classification of Functioning, Disability and Health (ICF). Results showed high usability, low cybersickness symptoms, and short-term improvements in attention, coordination, and engagement. The proposed framework demonstrates a scalable and accessible approach for inclusive sport training and cognitive-motor empowerment through immersive VR cycling.