As a result of the static and dynamic instabilities of a Powered-Two-Wheeler, the rider performs a highly demanding control task. Rider safety strongly depends on the individual abilities and skills of the rider. To improve the riders' skill level and reduce riding errors, safety trainings are well established. Additionally, safety systems and recently also advanced rider assistance systems help to avoid or mitigate accidents. While conventional rider training is limited to a small number of training scenarios in a controlled environment, safety systems are typically limited to specific situations (collision warning) or physical limits of the vehicle (ABS). We propose a method to identify riding errors based on a statistical rider model for cornering scenarios and estimate a personal rider skill score. The intention is to extend rider skill training beyond organized events and towards personal self-training. Automatic scoring of the cornering skill level has to take into account the high variability in speed and local curvature as well as the variety of different traffic situations that may be encountered during a ride. We suggest to split the complex driving task e.g. riding along a winding road, into smaller control tasks e.g. roll-into-corner, stable lean, and roll-out-of-corner and analyze them separately first, then their sequence and transitions. We evaluate various approaches based on Hidden Markov models that can split a complex task into smaller segments and show indicators for rider skill based on the best segmentation model.
Rider skill identification by probabilistic segmentation into motorcycle maneuver primitives
2016-11-01
676673 byte
Conference paper
Electronic Resource
English
Motorcycle rider skill assessment
TIBKAT | 1985
|Rider Skill Influences on Motorcycle Maneuvering
SAE Technical Papers | 1978
|Rider skill influences on motorcycle maneuvering
Automotive engineering | 1978
|R3 - A Model for Motorcycle Rider Skill Development
British Library Conference Proceedings | 2003
|