Research of object classification algorithm based on LIDAR for UGV

Typically, obstacle that UGV have to avoid can be divided into two different types, static and moving-obstacle. UGV should have an ability to separate static and moving obstacle. Because avoidance method for moving-obstacle is different from method for static-obstacle. The type of output data of LIDAR is cloud points data. When UGV separates moving obstacles, it is difficult to that trace it, calculating vectors by point unit. So we are going to simplify cloud points data into a specific object data, and judge the object's movement. Also, we can recognize the object's feature and dimension using processed data. It is just first step to recognize the environment with LIDAR. This research proposes the algorithm that classifies a lot of objects from LIDAR's point cloud data; The 1st step is the segmentation that is extracting process some specific points from LIDAR's cloud point data and is clustering step using specific points. Next step is the classification based on segmentation data. Via this process, we are able to obtain the object data. This research will be basis of recognition and avoidance algorithm for moving obstacles.