Research And Implementation Of Automatic Driving Lane-change Behavior Based On Predetermined Trajectory

Lane-change behaviour is a hot topic in current autonomous driving research and is one of the key technologies for advanced autonomous driving,which includes the lane-change trajectory planning module and the trajectory tracking module.The purpose of lane change trajectory planning is to plan a safe and reliable lane change trajectory that can be executed by the vehicle,and the trajectory planning determines whether the vehicle can successfully change lanes.The purpose of trajectory tracking is to control the vehicle to track the lane-change trajectory in real time,and the trajectory tracking module determines the actual driving effect of the vehicle.In this thesis,we design a quintic polynomial lane-change trajectory planning algorithm based on predefined trajectories,and use high precision GPS road information as predefined trajectories for actual road scenarios,which simplifies the construction of road maps in the process of lane-change trajectory planning.In addition,by designing the fuzzy controller which can adjust the sliding mode parameters,we can optimize the trajectory tracking effect of the traditional algorithm.Therefore,this thesis studies the lane change behavior of autonomous vehicles on straight roads.The main contents include the following aspects :(1)An experimental platform for self-driving vehicles was built based on electric mobility scooters.For the autonomous driving straight lane change scenario,the intelligent transformation plan of the mobility scooter was designed in conjunction with the vehicle characteristics.The work includes how to select and install the sensors,how to design the vehicle chassis and how to develop the core software control for the autonomous driving.(2)Design of lane changing trajectory planning algorithm based on predetermined trajectory.In this thesis,by establishing the quintic polynomial trajectory curve model,combined with the Frenet coordinate system and the GPS trajectory information of the predetermined road,the lane changing trajectory is decomposed into transverse trajectory and longitudinal trajectory,which simplifies the calculation of trajectory error in the planning process and realizes the lane changing trajectory planning of vehicles.(3)Design of fuzzy adaptive sliding mode trajectory tracking controller.By comparing the advantages and disadvantages of different lane change trajectory tracking algorithms,the sliding mode controller with anti-interference and good real-time performance is used to realize the lane change trajectory tracking of vehicles.Aiming at the problem that the sliding mode parameters of sliding mode control are difficult to determine,the performance of vehicle trajectory tracking is improved by building a fuzzy controller that adaptively adjusts the sliding mode parameters.(4)Algorithm simulation and road test.To prove the validity of the algorithm in this thesis,the simulation experiments were designed for the lane-change trajectory planning algorithm and trajectory tracking algorithm by using Py Charm and Matlab software respectively.In addition,the validity of the algorithm was verified on real roads by the autonomous vehicles.