基于伪谱拼接法的自动泊车路径规划

Automatic Parking Path Planning Based on Pseudo-Spectral Splicing Method

在局促环境下自动泊车路径规划是自动泊车研究的难点。基于Gauss伪谱法开展智能车入库的路径规划研究,首先使用P-Norm函数刻画出车库位置和车辆边界,然后根据车库和车辆位置添加边界约束,并融合车辆运动学约束构建泊车路径规划控制模型,将最优控制问题转化为非线性规划问题(NLP),运用序列二次规划法进行求解。仿真实验结果显示:Gauss伪谱法能对狭窄空间内多阶段泊车进行整体求解,证明了该方法的可行性。

Automatic parking path planning in cramped environment is the difficulty of automatic parking research. Based on Gauss pseudo-spectral method, this paper studies the path planning of intelligent vehicle warehousing. Firstly, the p-norm function is used to depict the garage location and the boundary of smart cars. Then the path constraint is added according to the boundary between the garage and he intelligent vehicle, and the kinematics constraint is fused. The optimal control model of parking path planning is established. The optimal control problem is transformed into nonlinear programming problem (NLP) and solved by sequential quadratic programming. The simulation results show that Gauss pseudo-spectral method can solve the complex parking path planning problem in stages as a whole and meet the requirements of automatic parking in narrow space, which proves the feasibility of this method.