Research On Lane Change Trajectory Planning Of Intelligent Vehicle Based On Machine Vision

With the rapid development of multi-sensor information fusion,modern control theory,artificial intelligence and large-scale integrated circuits advanced technologies,the intelligent vehicle has been become a research hotspot in the vehicle engineering field.Its target is to increase the vehicle's autonomous driving ability and driving security.The potential “manlike” performance of the intelligent vehicle is attained by research on road perception in complex environment and vehicle motion planning in dynamic scenes.Therefore,this paper combines The National Natural Science Foundation of China Major Research Project(51278514)and the National Key R&D Project “Study on key fundamental issues of perception,decision-making and control of intelligence electric vehicles”(2016YFB0100905),the lane detection technology,the front vehicle detection technology,the lane change decision method,and the lane change trajectory planning technology are studied in this paper.The main research work of this paper is as follows:(1)Research on road image processing and target detection method based on machine vision.Apply digital image processing technology for image preprocess based on the camera imaging model,aiming at the phenomenon of weak edge undetected,an edge point detection algorithm based on gradient amplitude and edge direction was proposed;and then,with the combination of lane geometry constraints and bidirectional scanning method,lane feature point was selected and extracted,using the least squares method to achieve lane feature point fitting;finally,based on the HOG feature of vehicle tail image and support vector machine to train the vehicle detection model,a variable window was used to scan road image to complete the vehicle detection.(2)Research on lane change decision method of intelligent vehicles based on data mining.According to the problem that the complex lane changing scenes,numerous influencing factors,and increased requirements for the accuracy of the lane changing models,the characteristics of lane changing were analyzed.Under the framework of intelligent vehicle decision model,driven by the next generation of simulation(NGSIM)driving trajectory data,the data of lane change trajectory was extracted,and the parameters of each influencing factor at the start of the lane change were tapped to reconstruct the lane change decision scene,and the lane change decision rules were established.Based on this,the lane change decision model based on Bayesian theory was established.(3)Research on lane change trajectory planning method based on collaborative optimization with multiple performance targets of intelligent vehicles.Based on the grey prediction and vehicle kinematics theory,the safety lane change domain was established by analyze the critical collision point of the self-vehicle and surrounding vehicles;and then,the polynomial is used to describe the dynamic characteristics of the lane changing process of vehicles under barrier-free conditions,and the problem of lane change trajectory under barrier constraints into the polynomial coefficient;at last,an optimal lane change trajectory was selected by considering the vehicle security,comfort,and economy.