合肥工业大学学报自科版

合肥工业大学学报(自然科学版)

JOURNAL OF HEFEI UNIVERSITY OF TECHNOLOGY (NATURAL SCIENCE)

主管：中华人民共和国教育部

主办：合肥工业大学

ISSN 1003-5060, CN 34-1083/N

基于粒子群-遗传算法的多约束平行泊车路径规划

下载全文

Multi-constraint parallel parking path planning based on particle swarm optimization-genetic algorithm

期刊信息

合肥工业大学（自然科学版），2026年3月，第49卷第3期：289-294

DOI: 10.3969/j.issn.1003-5060.2026.03.001

作者信息

尹晨晨 ¹，张冰战 ^2,3，康谷峰 ²，邱明明 ⁴

（1. 比亚迪汽车工业有限公司，广东深圳 518118；2. 合肥工业大学汽车与交通工程学院，安徽合肥 230009；3. 合肥工业大学安徽省数字化设计与制造重点实验室，安徽合肥 230009；4. 合肥工业大学机械工程学院，安徽合肥 230009）

摘要和关键词

摘要: 为满足多约束条件下汽车平行泊车的安全性、泊车效率以及车辆自身运动学要求,文章提出一种基于粒子群-遗传算法应用于平行泊车工况的路径规划算法。分析平行泊车过程,采用五次多项式曲线作为轨迹规划的基础,通过构建车辆运动学模型,分析泊车过程中运动学约束、碰撞约束和曲线端点约束,将车辆泊车路径规划问题转化为最优控制问题;采用粒子群-遗传算法进行求解,获得路径曲线的泊车起始点坐标,并计算路径曲线系数,以得到满足约束要求的路径曲线;基于模型预测算法进行泊车路径跟踪仿真,得到泊车过程中绝对误差最大为0.048 m,车身方位角最大误差为0.025 rad(约1.5°)。仿真结果表明:该算法能够较精确地跟踪优化后的五次多项式曲线路径,泊车起始点车身姿态与车位线平行,满足泊车路径端点要求;同时避免了与周围障碍物的碰撞,符合泊车安全性要求;前轮转角变化连续且曲线平滑,符合泊车过程汽车运动学要求。文章基于粒子群-遗传算法的路径规划算法可以为自动泊车提供有效参考路径。

关键词: 平行泊车；路径规划；自动泊车；粒子群-遗传算法；模型预测控制

Authors

YIN Chenchen ¹, ZHANG Bingzhan ^2,3, KANG Gufeng ², QIU Mingming ⁴

(1. BYD Auto Industry Co., Ltd., Shenzhen 518118, China; 2. School of Automobile and Traffic Engineering, Hefei University of Technology, Hefei 230009, China; 3. Anhui Key Laboratory of Digit Design and Manufacturing, Hefei University of Technology, Hefei 230009, China; 4. School of Mechanical Engineering, Hefei University of Technology, Hefei 230009, China)

Abstract and Keywords

Abstract: In order to meet the safety, parking efficiency and the vehicle's kinematics requirements of parallel parking under multiple constraints, a path planning algorithm for parallel parking based on particle swarm optimization-genetic algorithm is proposed. The parallel parking process is analyzed, and a quintic polynomial curve is adopted as the basis of trajectory planning, and the vehicle parking path planning problem is transformed into an optimal control problem by constructing a vehicle kinematics model and analyzing the kinematics constraints, collision constraints, and curve endpoint constraints in the parking process. The particle swarm optimization-genetic algorithm is used to solve the problem, the coordinates of the parking start point of the path curve are obtained, and the coefficients of the path curve are calculated to obtain the path curve that meets the constraint requirements. The model prediction algorithm based parking path tracking simulation shows that the absolute error in the parking process is 0.048 m at most, and the maximum error of body azimuth is 0.025 rad (about 1.5°). The results indicate that the algorithm can track the optimized path of the quintic polynomial curve. curve more accurately, and the attitude of the body at the starting point of the parking is parallel to the parking line, which meets the requirements of the endpoints of the parking path; at the same time, it avoids the collision with the surrounding obstacles, which meets the parking safety requirements; the front wheel angle change is continuous and the curve is smooth, which also meets the kinematics requirements of the vehicle in the parking process. The proposed algorithm can provide an effective reference path for automatic parking.

Keywords: parallel parking; path planning; automatic parking; particle swarm optimization-genetic algorithm; model predictive control

基金信息

国家自然科学基金资助项目（52172344）；中央高校基本科研业务费专项资金资助项目（PA2023GDSK0065）和芜湖市科技计划资助项目（2223jc-04）

后一篇

返回期刊目录