Research On Intelligent Perception Technology Of Autonomous Millimeter Wave Radar On Wet And Slippery Road Surface

Assisted driving and autonomous driving have now become an extremely important technological approach to upgrade the automobile industry.The automotive millimeterwave radar has been developed until today.Due to its all-weather high reliability and high precision detection ability of target motion characteristics,it has occupied a place in the automotive sensors.The research on the wet and slippery road surface sensing technology based on millimeter wave radar may provide more multi-dimensional information support for the future autonomous driving strategy.This thesis focuses on the new application of millimeter wave radar perception of road surface conditions,carries out electromagnetic scattering characteristics analysis and features extraction of road surface.A multi-feature fusion perception classification technology for wet and slippery road surface was proposed.Based on the measured data,experiments are conducted to verify that the perception of slippery road surface has obtained a high accuracy,which provides beneficial technical support for reliable automatic driving.The main work and contribution of this thesis are as follows:1.The road surface modeling method and electromagnetic scattering theory of random rough surfaces are analyzed,and the rough surfaces with different roughness and dielectric constant are modeled and electromagnetic scattering simulation is carried out,which provides theoretical support for the subsequent research on the pavement information acquisition of millimeter wave radar.2.The working principle of millimeter wave radar system and the road surface information measurement theory of LFMCW radar signal are studied.The radar echo signal processing is completed by referring to the theory,and the feature extraction and feature transformation methods of target road surface radar data are analyzed.3.A variety of classification algorithms are analyzed and their characteristics are compared.A kernel SVM classifier,based on the measured data,with optimized parameters of pavement multi-feature fusion is constructed.The established measured data set is used to train and optimize the classifier.4.the experimental design was carried out based on millimeter wave cascade experiment platform.The proposed feature fusion classification algorithm is tested and evaluated with the measured data set.The overall prediction accuracy of the system reaches 98%,which verifies the feasibility of the scheme: millimeter wave radar is used to sense slippery roads.