Research On Automatic Vertical Parking Path Planning Method Considering Starting Heading

With the rapid growth of automobile ownership,while bringing convenience to people’s lives,the problems of road congestion and parking difficulties are increasingly prominent.In order to solve the problem of parking difficulties,the automatic parking assistance system has emerged.Through sufficient research on relevant literature,it was found that the current parking path planning system still has problems such as high requirements for the initial pose,low path search efficiency,and poor path smoothness.To solve the above problems,a path planning method combining the feasible starting area of parking and the hybrid A* algorithm is proposed,and the path optimization strategy based on the quadratic programming method is adopted,and finally a reasonable path tracking controller is designed for tracking verification.Firstly,based on the characteristics of parking conditions and the Ackermann steering principle,a vehicle kinematic model suitable for parking conditions is established.In order to overcome the limitation of the collision constraint only applicable to the starting heading angle of zero in traditional vertical parking spaces,a collision constraint derivation method considering the starting heading angle is proposed.Based on the boundary conditions of the collision constraint,a feasible starting region for vertical parking is established,and a one-segment circular arc parking path and a three-segment circular arc parking path with initial heading angle are proposed in the feasible region that satisfies the collision constraint.For the region that does not satisfy the collision constraint,a path adjustment strategy based on the hybrid A* planning algorithm is used.The vehicle is first adjusted from the initial pose to the feasible region that satisfies the collision constraint,and then a one-segment or three-segment circular arc parking path is executed to park the vehicle.Through batch simulation experiments,a series of sampling points determined by a fixed step length within the feasible starting region are used as starting pose points,and compared with the direct use of the hybrid A* parking path planning method.The results show that the proposed path planning method can greatly reduce the dependence on the starting pose and has high real-time performance and robustness.Secondly,due to the initial path planning with sudden curvature changes and optimization space in path length,inspired by the reference line smoothing algorithm,a multi-objective function optimization method based on quadratic programming is designed.Obstacle avoidance constraints are added on the basis of the traditional line smoothing algorithm,so as to convert path optimization into a quadratic programming problem,optimizing the smoothness,geometric similarity and path length of the path.Simulation verification shows that the optimization effect is ideal.Then,an automatic parking path tracking controller is designed,with the longitudinal speed controller being a PID controller and the lateral controller being an LQR controller based on the kinematic model.Simulation verification shows that the controller has good control effect.Finally,a co-simulation model was established in Car Sim and Simulink platforms to verify the path tracking performance in different parking starting areas,and to further verify the robustness of the path planning algorithm.The simulation results show that the path planning algorithm and path optimization strategy designed in this paper can plan a smooth collisionfree path under each starting pose,and have high search efficiency.