Motion Estimation in Vehicle Rollover

摘要

Rollover detection requires an accurate knowledge of the vehicle's motion. Especially its angular velocity about the longitudinal axis and the angle relative to an horizontal plane are of major interest. Instead of using expensive gyroscopes formeasuring the angular velocity, a motion estimation using acceleration signals only is presented. An extended Kalman filter is implemented to guarantee optimal estimation results. The influence of different sampling rates is investigated, as low samplerates are required for a possible real time implementation. Orientation estimation relative to an inertial coordinate system is performed by recursively evaluating a rotation matrix. The usage of roughly quantized measurement signals is investigated tocope with the inadequate range of presently used accelerometers for safety systems.