Multi-Target Tracking Method For Collaborative RSUs

Intelligent Road Side Unit(RSU)is an important part of vehicle road cooperation system.It has the functions of communication,target detection and tracking.In the existing solution,each RSU can detect and track the target independently,and there is no cooperation between adjacent RSUs,resulting in insufficient information utilization and low tracking accuracy.Therefore,the research on multi-target tracking method based on RSU cooperation to improve the tracking accuracy of roadside sensors has theoretical significance and practical value.The main work of this thesis includes the following three aspects:The existing roadside lidar target detection relies on adjacent roadside sensors to make up for each other’s blind areas,resulting in intensive roadside sensor installation and low resource utilization.In order to solve the above problems,a target detection scheme of roadside lidar enhanced by blinding lidar is proposed in this thesis.The scheme combines roadside lidar and blinding radar into a whole,and uses blinding radar to compensate the bottom blind area of lidar.In target detection,firstly,the non ground points of two groups of radar point cloud data are extracted by region of interest division and ground plane fitting algorithm;Then,the non ground point cloud coordinates are registered and fused by iterative nearest point algorithm;Finally,the grid map is established,and the grid clustering method is used to cluster the point cloud to obtain the target information.This scheme can effectively expand the target detection range of a single RSU and reduce the number of sensors deployed by about47%。In order to solve the problem of low tracking accuracy caused by the independent target detection and tracking of each RSU,a multi-target tracking method based on RSU cooperation is proposed by using the information of adjacent RSUs.In this method,the RSU first receives the target trajectory information broadcast by the adjacent RSU as the potential target,and then measures and fuses the target detected by the roadside sensor and associates it with the existing target trajectory and potential target trajectory.The potential target trajectory on the correlation is regarded as the existing target trajectory,while the measurement on the non correlation is used for the start of new target tracking.Finally,the existing target trajectory is updated and broadcast through the tracking filter to realize RSU cooperative tracking.This method can effectively improve the target tracking performance,and the improvement accuracy for targets with different speeds is about 6% ～ 48%.According to the above algorithm and scheme,this thesis develops a multi-target tracking system with RSU cooperation,and carries out real vehicle experiments to verify the effectiveness of this method.