Research On Local Path Planning And Tracking Control Of Intelligent Vehicle

In the wake of booming of 5G wave and the continuous development of frontier technologies in term of computers,sensors,artificial intelligence,and cloud computing,China’s intelligent driving has entered a new era.Intelligent vehicles,as the development direction of the automobile industry for the time to come,represents the country’s scientific and technological strength.Thus,an increasing number of car companies,universities and research institutes are sparing no effort in the research and development relating with intelligent vehicles.Path planning and path tracking are most crucial in the research effort of intelligent vehicles.The quality of planned path and the accuracy of path tracking are the important standards to measure the intelligent degree of vehicles.Based on this fact,this thesis has taken the path planning and path tracking as its object.Following a full analysis on the pros and cons of each path planning and control algorithm,this research has selected the upgraded artificial potential field method(APF)and the optimized model predictive control method(MPC)in the planning module and the control module respectively.The major work are the following three aspects:(1)In order to address the local optimal problem of the algorithm,local minimum detection is adopted,meanwhile virtual sub-target is added.To solve the issue of unreachable target,the repulsive potential field function is improved to incorporate the relative distance between the intelligent vehicles and the target point.By doing so,the factor of distance can be employed to adjust the repulsive potential field.As for the limitation exerted by road to the path range of APF algorithm,the road constraint is taken into account,and the road repulsion field function is thus introduced and established.When it comes to the issue that the traditional APF method is unable to detect the motion information of obstacles,entailing potential collision of vehicles due to delayed retreat,the velocity potential field and acceleration potential field of obstacles are introduced,to make the artificial potential field more practical in dynamic obstacle avoidance.(2)Model predictive controller is designed to optimize its control parameters.In order to more accurately reflect the motion characteristics of intelligent vehicles,the three degree-of-freedom vehicle dynamics model and the magic formula tire model are established,from which,a discrete and linearized prediction model is obtained.For the purpose of ensuring the stability and comfort of intelligent vehicles during the process of path tracking,the tire sidesaw angle,the front wheel angle,and the sidesaw angle of the mass center are incorporated as the constraint factors,an objective function including the state quantity deviation and the control increment is established.In addition,multi-objective optimization regarding prediction time domain and control time domain is conducted with the tool of genetic algorithm,so as to obtain optimal time domain at different speeds.Besides,an optimized MPC controller is introduced to follow the expected path,by the means of lateral control of the vehicle through controlling the front wheel angle.(3)The path planning and tracking control modules on intelligent vehicle are subject to co-simulation to verify the effect of dynamic route planning and control.The APF method and MPC algorithm are programmed by using Matlab,and the co-simulation test platform composed of MATLAB and CARSIM is established.Following the above work,the simulation test on the path planning and tracking of intelligent vehicles,in the scenario of lane changing,overtaking and multi-obstacle is carried out to verify whether the algorithm is scientific and effective or not.The research results have showcased that the upgraded APF method can better meet the requirements of real time obstacles avoidance in terms of intelligent vehicles by its effective path planning.Whereas the optimized model prediction controller not only has minor tracking deviations,also good stability,making the steady tracking on reference path available.This thesis therefore,has provided a high reference value in terms of path planning and path tracking for intelligent vehicles.