基于卡尔曼滤波的激光雷达点云跟踪方法研究

Research on Lidar Point Cloud Tracking Method Based on Kalman Filtering

激光雷达点云目标跟踪是环境感知与定位的重要研究内容,具有广泛的应用前景。由于激光雷达存在探测范围有限、易受遮挡和散射条件影响等问题,无法保证跟踪过程中目标点云特征的完整性和正确性,跟踪目标丢失时有出现。为了提高跟踪过程中的鲁棒性,提出一种基于卡尔曼滤波的2D激光点云跟踪方法。该方法首先对聚类后的点云建立包围盒集合,其次对包围盒跟踪框建立卡尔曼滤波模型,将前一帧跟踪框最优估计值与当前帧包围盒,通过交并比方法进行关联性匹配,将关联性最大结果作为当前帧目标点云,并更新卡尔曼滤波模型参数,从而实现目标点云的实时跟踪。实验结果表明,所提方法提高了跟踪过程的鲁棒性和准确性。

Due to the limited detection range,and the vulnerability tolight occlusion and scattering,the integrity and correctness of target point cloud features can not always be guaranteedby lidar in the tracking process.Hence,target fails to be tracked sometimes.In order to improve the robustness ofthe tracking process,a laser point cloud tracking method based on Kalman filtering is proposed.Firstly,the bounding box set is established for the clustered point cloud,and then the Kalman filter model is used to develop the tracking framefor the bounding box.The optimal estimate of the tracking framein the previous frame and the bounding box in the current frame are matched by means of the intersection over union(IoU)method.Then,the maximum correlation result is taken as the target point cloud of the current frame,and the Kalman filtering model parameters are also updated at the same time.The target point cloud can be tracked in the real-time way.