Research And Implementation Of Automatic Driving Target Detection Based On FMCW Radar

In recent years,millimeter-wave radar has been widely used in the field of traffic detection due to its good detection ability in dark and occluded environments,and has become one of the necessary sensors for autonomous vehicles.By detecting the surrounding environment,the vehicle-mounted millimeter-wave radar provides the driver with blind spot vision and early warning of dangerous targets,thereby greatly reducing the occurrence of traffic accidents.Therefore,in the field of autonomous driving,the target detection and trajectory prediction capabilities of millimeter-wave radar are very important.In this thesis,the hardware circuit board is designed based on the 77 GHz radar sensor chip AWR1642,and the target detection algorithm and the trajectory extraction algorithm are optimized.Experiments show that the algorithm can realize target detection and trajectory extraction in complex environments.The main research contents of this thesis include the following three aspects:A target detection hardware board based on AWR1642 radar sensor is designed.It mainly includes the design of the power system of the radar board,the design of the AWR1642 and its peripheral circuits and interfaces,the layer design of the Printed Circuit Board(PCB)and the layout of each module.Finally,the test of board was completed,and the basic target detection algorithm was tested through the radar board,and it was found that there were many missed detections and false alarms.Aiming at the problems of missed detection and false alarms in the process of millimeter wave radar target detection,two-dimensional Fast Fourier Transform(2D-FFT)and beamforming algorithms are used to achieve high-precision range and angle measurement of radar sensors.On the basis of high-precision detection of target distance and angle,the Constant False Alarm Rate Detector(CFAR)detection algorithm is optimized to reduce missed detection and false alarms.Simulation and actual measurement show that the optimized CFAR detection algorithm can effectively improve the ability to distinguish adjacent targets in the road.The trajectory extraction algorithm is implemented.A clustering method combining reflection point density,reflection signal strength and target speed is proposed.Kalman filter is used to predict and generate prediction vector,and the similarity between the clustered feature vector and the prediction vector of the previous frame is used to match the target trajectory.Finally,the trajectory extraction of common moving objects on the road is realized.This algorithm can accurately extract the moving trajectories of various objects such as pedestrians,bicycles and cars.