Study On Vehicle Detection And Tracking Algorithm Based On Lidar

Environmental perception is a key link in the interaction between intelligent driving vehicles and environmental information.The randomness of the vehicle’s motion will reduce the accuracy of the intelligent driving vehicle’s perception of the environment in the driving scene.This thesis is based on the vehicle-mounted laser sensing system for vehicle detection and tracking technology.First,the vehicle-mounted laser sensing system which consists of lidar,integrated navigation system,and on-board computer is designed.The point cloud data is collected by lidar,and the point cloud data is corrected by the integrated navigation system,and the onboard computer is used for data processing.The GPS system time is used as the clock source to service the lidar,and the attitude of the lidar is calibrated in the integrated navigation coordinate system using the sampling and registration method,ensuring the time synchronization and spatial unity of the sensor data.Secondly,in view of the low accuracy of the existing vehicle detection methods for vehicle length,width and yaw angle detection,a bounding box fitting algorithm based on a priori model of vehicle point cloud is proposed.In order to reduce the calculation amount of the algorithm,the convex hull fitting algorithm is used to extract the boundary points of the vehicle target point cloud,and then different optimal bounding box fitting methods are selected to calculate the bounding box parameters according to the boundary points conforming to the L or I-shaped model.And the superiority of this algorithm is verified through experiments.Finally,in order to solve the problem of inaccurate vehicle tracking caused by the change of Lidar’s perspective,a vehicle tracking algorithm based on vehicle multi-parameter correlation is studied.The weighted Euclidean distance of the vehicle’s length,width,yaw angle,and position parameters is used as the measurement standard of the nearest neighbor data correlation algorithm,and the vehicle position measurement results are corrected according to the correlation results of the vehicle geometric parameters to improve vehicle tracking accuracy.At the same time,the effectiveness of the algorithm is verified through experiments.