Study On Point Cloud Imaging Technology For Vehicle-mounted Millimeter Wave Radar

The vigorous development of the new energy industry has promoted the development of the automotive industry towards electrification,while also accelerating the development of advanced driving assistance systems and autonomous driving technology.High performance sensing system is an important part of achieving the above goals.Millimeter wave radar has become an indispensable sensor in the sensing system due to its advantages of all-day,allweather,small size and low cost.It works together with other sensors such as laser radar and high-definition camera to provide the vehicle and engine system with information about the surrounding environment.Therefore,millimeter wave radar point cloud imaging has become a hot research topic.In this paper,the point cloud imaging process of millimeter wave radar is studied,so as to propose a set of millimeter wave radar point cloud imaging scheme with 3DFFT as the core.Specific work is as follows:(1)The traditional millimeter wave radar point cloud imaging technology is reviewed and analyzed,and on this basis,a set of millimeter wave radar point cloud imaging scheme based on 3D-FFT algorithm is proposed.In this scheme,the distance,velocity,angle and other information of the target can be obtained by FFT of radar echo data for three times,and then the point cloud image of the target can be obtained by coordinate mapping.(2)In order to solve the problem of poor accuracy of millimeter wave radar angle measurement,multi input multi output technology is introduced.This technology is used in combination with virtual aperture technology to achieve the purpose of using a small number of antenna array elements to obtain a larger antenna aperture,and significantly improve the angular resolution of radar.Numerical simulation experiments are carried out to study the influence of array number on the angle measurement accuracy of radar.The results show that the angle measurement accuracy of radar improves with the increase of array number.(3)In view of the poor clustering performance of the traditional DBSCAN clustering algorithm in the multi-objective non-uniform scene,an elliptical neighborhood was proposed to replace the circular neighborhood,and then the elliptical DBSCAN algorithm was developed.Numerical simulation is carried out to study the clustering performance of the algorithm in the multi-objective non-uniform scenario.The result shows that the clustering performance of the elliptic DBSCAN algorithm is better than that of the traditional DBSCAN algorithm in this scenario.(4)Aiming at the problems of multi-object masking and edge data undetectable in constant false alarm probability detection technology,a scheme combining two-dimensional ordered statistics constant false alarm probability detection algorithm and data flipping method is proposed.The simulation results show that the proposed scheme can maintain good detection results in two kinds of environments,namely uniform clutter and clutter edge.(5)The experimental platform was built based on the millimeter wave radar produced by Texas Instruments.Based on this platform,single-target,multi-target,static and dynamic experiments are carried out.The results show that the millimeter wave radar point cloud imaging algorithm with 3D-FFT algorithm as the core can realize the point cloud imaging of targets.At the same time,it was found that the number of point clouds formed by different targets in the azimuth direction varies greatly,and the degree of dispersion varies.