A single framework for vehicle roll, pitch, yaw estimation and obstacles detection by stereovision

This paper presents original framework and method for robust estimation of vehicle roll, pitch, yaw and obstacles detection using stereovision. The "v-disparity" algorithm for generic road obstacles detection, described and evaluated by Labayrade et al. (2002, 2003), is here extended to provide accurate and reliable values of the Euler angles of the vehicle. The algorithm can be divided in three main stages. The first one is designed for roll and pitch estimation. The second one is about obstacles detection and free road estimation. The third one concerns yaw estimation. Each stage is presented and experimental results are shown. In particular, the roll angle is estimated with a precision better than 0.2°.