Research On Target Location Estimation Based On Vehicle Dual Radar System

Based on the increasingly complex traffic environment,the intelligent and safety of vehicles has become the most concerned problem.In this thesis,driverless technology has gradually become one of the most potential directions in the field of intelligent technology.Unmanned driving is a technology that combines sensors with software algorithm and artificial intelligence technology,which have the functions of ranging,imaging and positioning.It realizes autonomous driving of vehicles through data transmission at multiple levels,such as perception level,path planning level,decision control level and execution level.Multi-sensor fusion is the key to realize this process,which directly affects the accuracy of vehicle environment perception.In the perception layer,an important function of sensor fusion is target tracking,even if the vehicle acquires real-time state information of the target in the detection area.In the state information acquired by sensors,the information of target position is particularly critical,which is directly related to the safety of vehicles.In this paper,a method of position estimation for unmanned vehicle is proposed,which is based on dual radar system and uses Unscented Kalman filter to predict and update the position of the target according to the labeled radar data.The main work of this paper can be summarized as follows:1.According to the characteristics and parameters of sensors,several kinds of radar commonly used at present are compared,and millimeter wave radar and lidar are selected.The process of radar data preprocessing is designed,the radar data format is analyzed,the data acquisition program is compiled,and the radar data preprocessing is completed by calculating the data and screening the effective data.2.According to the model structure of driving scene,the unmanned driving architecture of the experimental vehicle is designed.According to the functions to be achieved,the sensor layout scheme is designed.The scheme can effectively realize the target detection of obstacles in front of the vehicle.3.On the basis of the system structure and layout scheme,sensors and on-board computers are loaded into the vehicle;coordinate calibration of dual-radar system composed of ESR millimeter-wave radar and 4-line lidar is completed according to the installed radar position;and data processing scheme of this paper is designed based on dual-radar system.4.The radar measurement model and target motion model are studied and analyzed.A target position estimation algorithm based on vehicle dual-radar system and Unscented Kalman filter as sensor fusion method is proposed.The real radar data are collected at the experimental site to verify the algorithm.Experiments show that the dual radar system improves the measurement accuracy of the whole system,and the proposed algorithm effectively improves the accuracy and time efficiency of target position estimation of the dual radar system.