Adaptable, Rider-in-the-Loop and Road-Quality-Aware Model for E-bike Dynamics

Light electric vehicles like e-bikes are gaining popularity as a sustainable and accessible urban transportation mode. However, accurate dynamic models are required to optimize the rider experience, particularly in shared mobility scenarios. In fact, existing models are often identified considering fixed rider mass and road conditions, limiting their ability to provide real-time and personalized assistance effectively. This limitation also hinders the development of effective strategies for reducing rider fatigue. Therefore, we propose a novel rider-in-the-loop, road-quality-aware approach to dynamically model ebikes' behavior, enabling real-time adaptation to varying conditions. Specifically, our system quickly adapts the system dynamics to the specific rider and road conditions, thanks to dedicated parameters that are online tuned at the beginning of a ride, and to the road-quality information provided in real-time by a dedicated road quality machine-learning classifier. Experimental validation confirms the promising performance of the proposed approach, which paves the way for the design of new control strategies to promote light electric vehicle adoption.