Research On Autonomous Overtaking Maneuver Based On Relative Position Information

Recently,with the development of sensing,information processing and computation technology,the Intelligent Transportation System(ITS)has also developed rapidly which also brought huge economic benefits to the society.People show more interest in the feeling of comfort during their travel.Autonomous vehicles have been favored by everyone due to the ability of appropriate environmental awareness,wireless communications and control capabilities.The main areas of ITS include the prevention and mitigation of accidents,safety issues,improvement of energy and infrastructure utilization efficiency.Among them,safety issues have always been one of the hot issues of concern.It is reported that 1.2 million people die from traffic accidents each year,and it is expected that the number of traffic accidents will continue to increase by 2020 without taking measures,among all the accidents,traffic accidents caused by improper operation of overtaking account for 4-10%.Therefore,among the operations that may occur during the driving process,overtaking is one of the most challenging maneuvers due to its complicated operation.If the overtaking vehicle can obtain the driving state of the surrounding vehicles during the overtaking process,the collision accident will be significantly reduced.This thesis analyzes the driving characteristics of the vehicle without knowing the trajectory of the overtaken vehicle by using the relative position information obtained by radar,sensor,etc.,and this result is an important basis for the overtaking control of the overtaking vehicle,thus obtain a higher safety overtaking strategy.Specific research content mainly includes the following aspects:In this thesis,the relative position information such as distance,rangle and velocity obtained by radar is used to detect the abnormal driving behavior of the overtaken vehicle,which lays a foundation for the subsequent overtaking model.Due to the frequent changes of the relative position information,it is difficult to judge the driving characteristics of the vehicle.Therefore,the fuzzy inference system is used to fuse the information in order to obtain the abnormal driving analysis result of the target vehicle.The design of fuzzification,fuzzy rule base and defuzzification process are all based on the actual overtaking scene.In order to simplify the overtaking control strategy of the overtaking vehicle,the driving state of the target vehicle is discretized as ?safe? and ?dangerous?.Based on the abnormal driving behavior analysis of overtaken vehicle,the optimal and safe overtaking model of the overtaken vehicle is established.In this paper,the state of the whole overtaking process is described by the finite state machine.Combined with the driving state of the overtaken vehicle,the classic algorithm Q in reinforcement learning is used to complete the transition between states and complete the overtaking process.Through a series of simulation experiments,we verify that the proposed method can obtain a safe overtaking strategy without knowing the trajectory of the overtaken vehicle.