Real-time Detecting And Tracking Of Moving Objects Using 3D Lidar

The research of the autonomous vehicle is one of the most important branches of the artificial intelligent. The research of moving objects detection and tracking we presented in this paper has fairly important meaning for the safely and accuracy driving of the autonomous vehicle. We present an algorithm for obstacle detection, and combined with the data association and Kalman filter, accomplished moving objects tracking. The main work of paper includes:1) For obstacle detection, it is hard to process efficiently under large amount of data. Our approach is first convert the Lidar data into a grid map using the constraint of road lane information from the image, then use region labeling and template matching for clustering and feature extracting, and finally use box model to represent these obstacles.2)For data assciation, in order to get the accurate matching information from multiple frames, we use multiple hypothesis tracking. Using this method, we can avoid the situation of false alarm or miss matching.3)For state estimation, in order to provide accurate moving information of tracked objects, we use Kalman filter to estimate every moving objects'situation. Then provided them to the planning module.