Research On Target Detection And Tracking Algorithms For Vehicle Mounted Millimeter Wave Radar

As a key component of the perception layer of advanced driving assistance systems,millimeter-wave radar has attracted extensive attention in recent years due to its outstanding advantages such as small size,low cost,and all-weather.Vehicle-mounted millimeter-wave radar obtains the information of the detection scene by signal processing the received echoes,and subsequent advanced driving assistance functions need to be realized based on the scene information.Therefore,the study of vehicle-mounted millimeter-wave radar signal processing algorithms has important theoretical significance and application value.This thesis mainly focuses on the target detection and tracking link in the signal processing process,and carries out the following research:Firstly,the variable index constant false alarm detection VI-CFAR algorithm with the adaptive characteristic of the detection background is studied.Aiming at the problem that the detection performance of traditional VI-CFAR drops sharply when there are interference targets in the left and right reference sliding window,based on the idea of truncation statistics,outliers are removed by adaptive iteration on the left and right reference sliding window samples;and the truncation The parameters of the final data are estimated to obtain the corresponding detection threshold to improve the detection performance of VI-CFAR in dense multi-target environments.In order to further reduce the time complexity of the algorithm to meet the real-time requirements of vehicle radar signal processing,based on the idea of replacement average,the detected target is replaced by the corresponding noise power estimation value,and an improved VI-CFAR algorithm with lower computational complexity is proposed.The simulation results show that the detection performance of the two improved algorithms is better than that of the traditional VI-CFAR algorithm when there are multiple interference targets on one side or both sides of the reference sliding window,and they can better adapt to the vehiclemounted dense multi-target detection characteristics.Then,the data association algorithm in target tracking is studied.In view of the fact that the traditional data association algorithm only uses the measurement position information,it is easy to make an association error in the case of target trajectory crossing or dense trajectory,and then the problem of trajectory exchange or loss occurs.Combined with the millimeter wave radar,the characteristics of the target speed information can be obtained.The speed information is added in the construction process of the data association cost matrix in order to improve the stability of target tracking.In the case that the traditional algorithm is difficult to correctly match the trajectory and the corresponding measurement only by using the position information,the simulation experiment results show that the improved algorithm can effectively improve the accuracy of the correlation between measurement and trajectory.Finally,based on TI’s AWR2243 cascaded radar measured data platform,using the collected measured data,the improved target detection and tracking algorithm proposed above was tested and verified.The experimental results show the effectiveness of the two improved VI-CFAR algorithms in actual scenarios;the improved data association algorithm can also improve the stability of the tracking process.