Driver Activity Recognition from Sitting Postures

Abstract

Automatically adapting vehicle settings due to activities of the driver has crucial impact to safety in road traffic, intelligent driver assistance and vehicle control systems as well as driving comfort. Distraction continues to increase by reason of technology growth (classical office- and home electronics like entertainment and navigation systems, televisions, Email and internet-clients, multimedia player, on-board computers, multi-function dashboards, game consoles, etc. are now built-in in cars). We study the possibilities of identifying typical drivers activities inside the car by evaluating pressure patterns collected from force sensor arrays, invisibly and unobtrusively integrated into the fabric of the driver seat and backrest. Along this key task, algorithms are developed for sitting posture recognition and empirical studies are performed to assess the reliability of activity recognition from sitting postures.