Research On High Resolution Road Environmental Perception Algorithm Based On 77GHz Automotive Radar

In recent years,autonomous driving is gradually moving from theoretical research to engineering reality.Sensors such as automotive radar and lidar,ultrasonic radar,and cameras form the core of perception in autonomous vehicles and advanced driver assistance systems(ADAS).Automotive radar systems are responsible for detecting the position and speed of objects and obstacles relative to the vehicle.With the development of millimeter wave semiconductor technology,the development of signal processing technology plays a key role in automotive radar systems.Various signal processing technologies are being developed day by day to provide higher resolution and estimated performance in measurement dimensions such as range,azimuth,and velocity of vehicles around the target.The research content of this article covers many aspects of automotive radar signal processing and data processing technology,including waveform design,array structure,detection algorithm,tracking algorithm,correlation algorithm and processing methods in complex environments,and dual radar fusion to improve radar detection performance Detection method.In order to realize the stable detection and tracking of the target in the scene by the automotive radar,it is necessary to carry out basic signal processing and subsequent data processing for the acquired original echo data.Based on the detailed analysis of the development of MMW radar detection technology at home and abroad,this paper uses awr1642 For the 77 GHz short-range automotive radar,firstly,the signal processing flow of automotive radar is introduced,and the method of improving azimuth resolution of the virtual array synthesis technology of MIMO system is studied theoretically and experimentally.On this basis,the speed expansion algorithm of improving the maximum range of velocity measurement is studied by using the array phase relationship.In the aspect of target detection,the spectrum bending of range-doppler(RD)spectrum caused by the coupling of azimuth and doppler when the radar is placed on the moving platform is studied.Based on the RD spectrum,the cross window CFAR algorithm of Rayleigh,exponential and Weibull clutter models is studied.Based on the output of the target detection,the target clustering algorithm is studied,and the DBSCAN algorithm based on the density clustering is applied to extract the location,speed and other information of pedestrian and vehicle targets.In the aspect of target tracking,a state equation suitable for the state estimation of road vehicles in the geodetic coordinate system is constructed,and the joint probability data association(JPDA)algorithm is used to carry out the association filtering for the targets in the dense multi-target scene,and the detection tracking fusion algorithm is studied,which improves the tracking track continuity of the target with weak echo energy.Finally,two forward-looking radars are simulated,the geometric model of the forward-looking target is established,the variables related to the real speed dimension of the target are constructed to noncoherent accumulate the RD spectrum,and the range-velocity-azimuth spectrum(RVA spectrum)fusion is carried out by using the different perspectives of the two radars relative to the target,so as to improve the detection accuracy of the target.