Development of Localization Technology with Accumulation of Recognition Result

Abstract

We developed a self-localization algorithm estimated with high accuracy and robustness. The cameras and LRFs mounted to the vehicle recognize the lane lines and the curb stones around the vehicle and the recognition results are accumulated based on the velocity and yaw rate, as a result the recognition results of wide area are produced. Self-pose is estimated by matching them to the map with the position date of the lane lines and the curb stones.