摘要：AbstractWe have developed the three dimensional light detection and ranging (3D-LiDAR) which has wide field of view in the vertical direction suitable for driving autonomously in urban areas and crowds. In this paper, we propose a method to recognize dynamic obstacles from motion of objects without using shape information. We classify point clouds as obstacles from the distance relationship between point to point, observe the movement of obstacles by tracking their centers of gravity. For creating precise map in the dynamic environment, we remove the dynamic objects from the point-clouds. The characteristics of wide field-of-view of the LiDAR makes localization robust. Experiments at Meiji University Ikuta campus showed the usefulness of this 3D-LiDAR and this method.
关键词：KeywordsRecognitionObstacle detectionAutonomous mobile robotsRobust estimationNavigation system
