Unsupervised 3D Point Cloud Anti-Occlusion Multi-Object Tracking

Occluded sensing data can lead to wrong detection on object classes and sizes. This work proposes an unsupervised occlusion resilience tracking algorithm for point cloud data. In the data association stage, the object detection information and point cloud features are used to compensate motion prediction errors caused by occlusion. The features of the point cloud are extracted using Normal Distributions Transformation, which is designed based on probabilistic statistical rules to achieve unsupervised point cloud feature extraction and similarity comparison. The proposed method also improves the existing tracking algorithms, to trade the output quality with the maximum allowable occlusion time intervals of the object, namely tracking robustness. The proposed algorithm is evaluated on two datasets: Wayside (collected by stationary LiDAR) and KITTI (collected by mobile LiDAR). On Wayside dataset, using optimal detecter under occlusion scenarios for tracking achieves an MOTA of 95.81 % without any ID switches, significantly outperforming AB3DMOT. On KITTI validation set, compared to the CONF_MOT algorithm, the MOTA increases from 69.09 % to 71.11 %, and the number of ID switches are reduced for up to 50 %.