Research On Path Planning And Path Tracking Of Overtaking Vehicles In Complex Scenarios

The automobile industry has developed rapidly due to people’s reliance on convenient and fast ways to travel.At the same time,autonomous driving cars have become a hot topic in the industry to address issues such as traffic congestion and driving safety.Lane changing and overtaking,as a common driving behavior,often relies on the driver’s visual field observation and experience to judge.If the driver is in poor condition,it will cause problems such as lane change delay and rear-end collision.Therefore,this thesis focuses on the research of vehicle overtaking path planning and path tracking methods in complex scenarios.The main content of this thesis is as follows:Firstly,the driving situation of obstacle vehicles in the adjacent lanes to the ego lane is considered,and a complex overtaking scene is defined.Then,the reasons for lane changing and overtaking are analyzed in a certain typical structured road,and a lane changing safety distance model and a vehicle safety model are constructed.Based on the two models and using fuzzy control theory,a lane changing decision model that ensures traffic efficiency and lane changing safety is established,allowing the ego vehicle to switch reasonably between acceleration,following,and lane changing modes.The safe space for the ego vehicle to change lanes is constructed based on the principle of convex approximation for obstacle avoidance,limiting the selection range of the optimal lane changing trajectory.Finally,a brief introduction of commonly used coordinate systems for vehicles is given,and the Frenet coordinate system is introduced to decouple the trajectory planning problem and reduce its complexity.This thesis uses the ideas of "trajectory point and speed decoupling" and "initial planning+ quadratic optimization" for trajectory planning.Firstly,based on the capsule-shaped safety model,an additional safety model for vehicles is constructed to ensure the high-speed driving safety of the ego vehicle and obstacle vehicles.For trajectory point planning,sampling points are set in the safe space for lane changing,and different sampling areas are divided according to the number of sampling points.The distance between the ego vehicle and the obstacle vehicle is used to form a transition between them.Secondly,to prevent interference from obstacle vehicles and save computation resources,dynamic and static obstacle models are established based on the motion status of the obstacle vehicles,and the sampling points in the area occupied by the model are discarded.Then,the five-order polynomial curve is used to traverse the sampling points and obtain the rough trajectory cluster.The optimal trajectory is filtered out through the trajectory evaluation function,and the optimal lane-changing trajectory is obtained through quadratic optimization.In addition,standard trajectory cost values are set for different stages of lane changing to ensure the trajectory transition between different stages of overtaking.To address the issue of timely response of the vehicle in a complex environment,the execution cycle of the trajectory planner is set to ensure that the planned trajectory can be updated in real-time.The speed planning method is similar to the above-described approach.This thesis studies the trajectory tracking algorithm and establishes lateral trajectory tracking controller and speed tracking controller separately.The lateral trajectory tracking uses the model predictive control method,which builds a trajectory error model based on the threedegree-of-freedom vehicle dynamic model and linearizes it to predict the future state of the vehicle.Finally,the optimal steering angle of the vehicle is solved by constructing the constraint function and expected function.The longitudinal control uses the position-velocity dual PID control method,which sets up PID controllers for both positional error and velocity error,and constructs a non-parametric vehicle dynamic model.By inputting the desired acceleration,the throttle opening or braking pressure is outputted to control the speed,imitating human operation.Finally,in order to verify the stability of the designed algorithm,different lane-changing scenarios were simulated and tested through a joint simulation platform.The results showed that the trajectory planning algorithm was able to plan the optimal overtaking trajectory in real-time for some complex driving scenarios.The trajectory tracking method can stably track the trajectory with high accuracy.The combination of the two methods can successfully implement the vehicle’s lane-changing overtaking behavior.