Research On Target Tracking And Conflict Warning Strategy And Algorithm Based On Multi-Sensor Fusion

At present,as a convenient means of transportation,the car ownership is increasing rapidly every year.But with the increase of traffic accidents,people's lives and property suffer huge losses.Therefore,vehicle safety assistant driving technology an d driverless technology can well avoid traffic accidents caused by drivers.In this paper,based on the Provincial Natural Science Foundation,the target tracking in the vehicle driving environment is carried out based on the radar and Mobileye fusion algo rithm,and the conflict risk assessment algorithm and conflict early warning strategy are studied.Because of the problem of high false alarm rate in the data of radar detection,this paper puts forward the method of effective life cycle test after the effective target is preliminarily selected,which can judge the continuity of motion parameters in the adjacent period of the radar detection target and the continuous tracking of multi frame data,so as to make the effectiveness and stability of radar detection target The qualitative analysis has been greatly improved.The Kalman weighted fusion algorithm is used to fuse the sensor data after data association,and the interactive multi model algorithm is used to track the target,which avoids the problem of insufficient accuracy of single model algorithm.In view of a large number of research on single early warning strategies only considering braking or lane changing,this paper considers a variety of early warning strategies,and studies the early warning strategies with four early warning strategies: driving at the current state,decelerating at a certain deceleration,emergency braking,lane changing driving,and judges the switching time of the four early warning strategies,so that the four early warning strategies can be based on The corresponding matching of judgment conditions is more suitable for solving traffic conflicts in real traffic environment.Aiming at the problem that only one conflict risk assessment algorithm is used,so that the conflict risk assessment can not meet the requirements of high efficiency and accuracy at the same time,this paper compares four conflict risk assessment algorithms: vehicle state based,driving intention based,virtual force field based and vector histogram based In this paper,a real-time and accurate conflict risk assessment and early warning strategy selection system is formed by combining the conflict risk assessment algorithm based on vehicle status with the conflict risk assessment algorithm based on vector histogram.In addition,the method of fusion Kalman filter is adopted to solve the problem that the conflict risk assessment algorithm based on vector histogram cannot predict.Aiming at the target tracking algorithm of multi-sensor fusion and the early-warning strategy selection system based on the conflict risk assessment,this paper refits the unmanned driving experiment platform based on Harvard H7,and verifies the research content of this paper,realizes the target tracking stability,and the conflict risk assessment under different states is accurate,which basically meets the stable early-warning policy selection.