Driving-Style Assessment from a Motion Sickness Perspective Based on Machine Learning Techniques

Abstract

Ride comfort improvement in driving scenarios is gaining traction as a research topic. This work presents a direct methodology that utilizes measured car signals and combines data processing techniques and machine learning algorithms in order to identify driver actions that negatively affect passenger motion sickness. The obtained clustering models identify distinct driving patterns and associate them with the motion sickness levels suffered by the passenger, allowing a comfort-based driving recommendation system that reduces it. The designed and validated methodology shows satisfactory results, achieving (from a real datasheet) trained models that identify diverse interpretable clusters, while also shedding light on driving pattern differences. Therefore, a recommendation system to improve passenger motion sickness is proposed.