基于模型预测控制的平行泊车路径规划研究

Research on Parallel Parking Path Planning Based on Model Predictive Control

针对自动泊车路径规划研究较少采用模型预测控制(MPC)路径规划方法的现状,本文结合MPC和平行泊车路径规划的特点提出解决方案。该方案首先以单车运动学方程为基础模型、泊车起点与目标终点间的直线作为参考路径,经多步推导得到预测表达式;然后通过向量积方法建立避障约束,以车辆车速和转向的极值作为控制量约束;最后定义目标函数并转化为标准二次型。方案通过Matlab仿真不同泊车起点坐标、不同车位尺寸、航向角是否方正等工况,并从有效避障、路径平滑、位姿误差小和航向角连续等方面分析了其可行性。

Aimming at the current problem that model predictive control(MPC)path planning method is rarely used in the research field of automatic parking path planning,this paper puts forward a solution combining the characteristics of MPC and parallel parking path planning.Firstly,the scheme is based on the kinematic equation of the bicycle,and the reference path of straight line between the parking starting point and the target end point.Then,the obstacle avoidance constraints are established by vector product method,and the extreme values of vehicle speed and steering are taken as the control constraints.Finally,the objective function is defined and transformed into the standard quadratic form.The scheme is simulated by Matlab with different starting point coordinates,different parking space sizes and different heading angle,and its feasibility is analyzed from the dimensions of effective obstacle avoidance,smooth path,small pose error and continuous heading angle.