Autonomous Vehicle Motion Planning Based On Model Prediction In The Medium And High Speed Overtaking Scene

Autonomous driving technology is an important part of Smart Transportation and Smart City,and it will have great significance for human production and life.Among modules of autonomous driving technology,the motion planning module plays an integral role.It is responsible for processing the information of the sensing system,positioning system,and prediction system in real time,then it plans a safe and comfortable driving trajectory for the control module.In the medium-high-speed scene,the traffic situation around the car changes rapidly,and how to plan a safe and comfortable trajectory in real time is full of challenges.To this end,this article focuses on the unmanned vehicle motion planning in the medium and high speed overtaking scene.The motion planning problem in the medium and high speed overtaking scene has the following special characteristics: the vehicle is driving at high speed,and its motion state needs to be accurately modeled;the surrounding traffic conditions are changing rapidly,and the collision-free constraint conditions related to obstacle vehicles at high speed are difficult establishment;multi-constrained and non-linear optimization problems have high computational complexity and are prone to failure.Existing parametric curve methods,search methods,numerical optimization methods,and learning methods usually use particle models,kinematic models,or smooth curves to describe the vehicle’s motion state.Such modeling methods cannot accurately describe the vehicle’s motion at high speeds.State,the planned trajectory control module is difficult to track,and it is easy to cause the vehicle to flick and roll.There are also shortcomings in the environment modeling method of virtualizing surrounding vehicles into multiple continuous static obstacles.When the traffic flow density increases,such a processing method makes the driveable area narrow,resulting in trajectory planning failure.Aiming at the particularity of medium and high speed overtaking motion planning problems and the shortcomings of existing methods,this study models the motion planning problem into a numerical optimization model based on optimization theory.Perform force analysis on the vehicle,use the dynamic model to describe the motion state of the vehicle at high speed,establish the anti-skid constraint based on the maximum slip angle constraint,and establish the anti-roll restraint based on the maximum zero moment point offset constraint.The motion state at multiple moments is estimated to construct a collision-free constraint.In order to solve the problems of high complexity and poor stability caused by the introduction of vehicle dynamics models,this study uses a completely discrete numerical optimization method to solve the problem.The state variables and control variables of the system are simultaneously used as free variables to optimize the solution.Numerical stability.Aiming at non-linear,multiconstrained numerical optimization problems with high computational complexity and long time-consuming problems,this study classifies traffic scenarios and constructs a database in advance to solve the optimal solution corresponding to each scenario.Matching gets the initial value of the optimization iteration,which reduces the online optimization time.To verify the effectiveness of the proposed model-based motion planning method,a vehicle dynamics simulation platform based on Simulink,ROS,and Carsim was constructed in this study.By comparing the EM Planner motion planning algorithms in Baidu’s Apollo open source platform,under the same high-speed overtaking scenario,the proposed trajectories planning method makes the vehicle’s lateral load transfer rate and centroid side deflection angle smaller,safer and more comfortable.In order to verify the efficiency of the proposed optimized hot-start method based on the database matching method,we worked with the "0 initial value method","current state initial value method","RRT * initial value method","A * initial value method".By comparison,it can be seen that the initial value hot start method proposed in this study has the smallest total time.