汽车工程 ›› 2023, Vol. 45 ›› Issue (7): 1163-1173.doi: 10.19562/j.chinasae.qcgc.2023.07.007

所属专题: 智能网联汽车技术专题-规划&决策2023年

• 专题:汽车智能化关键技术 • 上一篇    下一篇

阶段约束下Gauss配点离散化平行车位自动泊车轨迹规划

刘平1,陈卓1,刘明杰1,3,朴昌浩1(),Soohyun Jang2,万凯林3   

  1. 1.重庆邮电大学自动化学院,重庆 400065
    2.韩国电子技术研究所,Korea Seongnam 13230
    3.重庆长安汽车股份有限公司,重庆 400023
  • 收稿日期:2022-12-27 修回日期:2023-02-06 出版日期:2023-07-25 发布日期:2023-07-25
  • 通讯作者: 朴昌浩 E-mail:piaoch@cqupt.edu.cn
  • 基金资助:
    国家重点研发计划项目(2022YFE0101000);重庆市自然科学基金面上项目(CTSB2022NSCQ-MSX0355)

Gauss Allocation Points Parameterization Parallel Automatic Parking Trajectory Planning for Vehicle Under Multi-Stage Constraints

Ping Liu1,Zhuo Chen1,Mingjie Liu1,3,Changhao Piao1(),Soohyun Jang2,Kailin Wan3   

  1. 1.College of Automation,Chongqing University of Posts and Telecommunications,Chongqing 400065
    2.Korea Electronics Technology Institute,Korea Seongnam 13230
    3.Chongqing Changan Automobile Co. ,Ltd. ,Chongqing 400023
  • Received:2022-12-27 Revised:2023-02-06 Online:2023-07-25 Published:2023-07-25
  • Contact: Changhao Piao E-mail:piaoch@cqupt.edu.cn

摘要:

针对平行车位自动泊车高精度轨迹规划,提出一种联立多阶段划分和非均匀Gauss配点参数化的轨迹优化算法。首先,结合车辆泊车动力学方程和约束条件建立了平行自动泊车数学模型;然后,根据泊车过程,提出将泊车过程划分为泊车起动、接近车位、进入车位、姿态调整、泊车落位5个阶段,并建立各阶段对应的不等式约束条件;进一步,在伪谱法算法框架下,结合阶段约束给出了多阶段局部高斯配点策略,以此实现分阶段独立配点离散,从而提高泊车轨迹规划的精准度和适应性;最后,在通用小型汽车模型上在长短车位上进行仿真测试,验证提出方法的正确性和适应性。结果显示,相较于分段区域Gauss伪谱法,改进方法在获得平滑泊车轨迹的同时泊车时间平均减少2.976 s,时间性能提升21.7%,表明了所提算法的有效性。

关键词: 自动泊车, 平行车位, 轨迹优化, 阶段划分, 高斯配点参数化

Abstract:

A trajectory optimization algorithm combining multi-stage division with non-uniform Gauss collocation parameterization is proposed for vehicle high-precision trajectory planning of parallel parking space. Firstly, a mathematical model of parallel automatic parking is established based on the dynamic equation of vehicle parking and constraint conditions. Then, according to the parking process, it is proposed to divide the parking process into five parking stages including parking start, parking space approaching, parking space entering, parking position adjustment and parking landing and corresponding inequality constraints are established for each stage. Furthermore, multiple stage local Gauss discrete strategy is proposed under pseudo-spectral approach frame to achieve independent allocation in each stage so as to improve the precision and adaptivity of trajectory planning. Finally, simulation tests are carried out on a general car model for long and short parking space to verify the performance of the proposed method in accuracy and adaptivity. The test results show that the proposed method can efficiently obtain smooth parking trajectory and averagely decrease parking time by 2.976 s when compared with the piecewise Gaussian pseudo-spectral method, with 21.7% improvement of parking time performance, which indicates the effectiveness of the proposed algorithm.

Key words: automatic parking, parallel parking space, trajectory optimization, stage division, Gauss allocation parameterization