无人驾驶公交车基于循迹误差观测和目标测量误差观测的避障路径规划算法

doi:10.3969/j.issn.1674-8484.2024.04.015

摘要/Abstract

摘要：

为了提高无人驾驶公交车的避障安全，针对循迹控制不确定性和目标测量不确定性导致地碰撞风险问题，提出了一种基于循迹误差观测和目标测量误差观测的避障路径规划算法。首先利用均值方差模型描述循迹误差，并引入Gaussian 过程回归量化目标测量误差。随后，建立了目标横向空间概率分布模型，为安全横向偏移值的求解提供理论支持。最后，通过Bézier路径模型实现避障路径规划, 并建立综合评价体系对避障规划性能进行评价。在仿真环境下，探究了循迹误差以及测量误差等不确定因素对避障品质的影响, 并在天津大学无人驾驶智能公交的测试道路上进行了实车数据仿真验证，对3种算法的效果进行评价。结果表明：实际行驶中，避障安全性提高了17.61%，具有较好的鲁棒性，提出的算法平均耗时7.43 ms，具有良好的实时性。

关键词: 无人驾驶公交车, 避障路径规划, 循迹误差观测, 目标测量误差观测

Abstract:

An obstacle avoidance path planning algorithm based on tracking error observation and target measurement error observation was proposed to enhance the obstacle avoidance safety of autonomous buses, aiming at the risk of collision caused by the uncertainty of tracking control and target measurement. The mean-variance model was utilized to describe the tracking error, and the Gaussian process regression was introduced to quantify target measurement error. A probability distribution model for the lateral spatial position was established, providing theoretical support for solving the safe lateral offset value. An obstacle avoidance path planning was implemented through the Bézier path model, and a comprehensive evaluation system was established to assess the performance of the obstacle avoidance planning. In the simulation environment, the influence of uncertain factors such as tracking error and measurement error on obstacle avoidance quality was explored, and the real vehicle data simulation verification was carried out on the test road of unmanned intelligent bus in Tianjin University, and the effect of the three algorithms was evaluated. The results show that the obstacle avoidance safety is improved by 17.61% in actual driving, and it has good robustness. The proposed average time of the algorithm is 7.43 ms, which has good real-time performance.

Key words: autonomous buses, avoidance path planning, tracking error observation, target measurement error observation

中图分类号:

U461.91

李玉龙, 谢辉, 宋康. 无人驾驶公交车基于循迹误差观测和目标测量误差观测的避障路径规划算法[J]. 汽车安全与节能学报, 2024, 15(4): 579-590.

LI Yulong, XIE Hui, SONG Kang. An obstacle avoidance path planning algorithm for autonomous buses based on tracking error observation and target measurement error observation[J]. Journal of Automotive Safety and Energy, 2024, 15(4): 579-590.

图/表 20

参考文献 15

[1]	交通管理局. 全国机动车保有量达4.17亿辆驾驶人超过5亿人[R/OL]. (2023-01-11). www.gov.cn/xinwen/.
	TRAFFIC Control Bureau. The number of motor vehicles in the country has reached 417 million and the number of drivers has exceeded 500 million[R]. (2023-01-11). www.gov.cn/xinwen/. (in Chinese)
[2]	Omveer S, Sahoo C N, Puhan B N. Recent advances in motion and behavior planning techniques for software architecture of autonomous vehicles: A state-of-the-art survey[J]. Eng Appl Artif Intel, 2021, 101: 104211.
[3]	鲍久圣, 张牧野, 葛世荣, 等. 基于改进A*和人工势场算法的无轨胶轮车井下无人驾驶路径规划[J]. 煤炭学报, 2022, 47(3): 1347-1360.
	BAO Jiusheng, ZHANG Muye, GE Shirong, et al. Underground driverless path planning of trackless rubber tyred vehicle based on improved A* and artificial potential field algorithm[J]. J Chin Coal Soc, 2022, 47(3): 1347-1360. (in Chinese)
[4]	SHENG Weitian, LI Bai, ZHONG Xiang. Autonomous parking trajectory planning with tiny passages: A combination of multistage hybrid A-Star algorithm and numerical optimal control[J]. IEEE Access, 2021, 9: 102801-102810.
[5]	ZHONG Xunyu, TIAN Jun, HU Huosheng, et al. Hybrid path planning based on safe A* algorithm and adaptive window approach for mobile robot in large-scale dynamic environment[J]. J Intel Robot Syst, 2020, 99: 65-77.
[6]	Lim M, Lee S, Sunwoo M, et al. Hybrid Trajectory Planning for autonomous driving in on-road dynamic scenarios[J]. IEEE Trans Intel Transport Syst, 2021, 22(1): 341-355.
[7]	YANG Da, ZHENG Shiyu, WEN Cheng, et al. A dynamic lane-changing trajectory planning model for automated vehicles[J]. Transport Res Part C, 2018, 95: 228-247.
[8]	CHEN Jianyu, ZHAN Wei, Tomizuka M. Autonomous driving motion planning with constrained iterative LQR[J]. IEEE Trans Intel Vehi, 2019, 4(2): 244-254.
[9]	ZHANG Xinglong, ZHANG Wenxin, ZHAO Youqun, et al. Personalized motion planning and tracking control for autonomous vehicles obstacle avoidance[J]. IEEE Trans Vehi Tech, 2022, 71(5): 4733-4747.
[10]	周扬, 谢辉, 肖蓬勃, 等. 基于主动优化的无人驾驶客车实时性运动规划算法[J]. 汽车安全与节能学报, 2020, 11(4): 476-486.
	ZHOU Yang, XIE Hui, XIAO Pengbo, et al. Real-time motion planning algorithm for autonomous bus based on initiative optimization[J]. J Automo Safe Energ, 2020, 11(4): 476-486. (in Chinese)
[11]	李娜. GEE框架下纵向数据均值-方差联合模型的研究[D]. 昆明: 云南财经大学, 2021.
	LI Na. Research on joint mean-covariance modelling for longitudinal data within the framework of generalised estimating equations[D]. Kunming: Yunnan University of Finance and Economics, 2021. (in Chinese)
[12]	陈雪梅, 李梦溪, 王子嘉, 等. 无人驾驶车辆城市交叉口周边车辆轨迹预测[J]. 汽车工程学报, 2021, 11(4): 235-242.
	CHEN Xuemei, LI Mengxi, WANG Zijia, et al. Trajectory Prediction of Surrounding Vehicles for Unmanned Vehicle at Urban Intersections[J]. Qiche Gongcheng, 2021, 11(4): 235-242. (in Chinese)
[13]	Durakli Z, Nabiyev V. A new approach based on Bezier curves to solve path planning problems for mobile robots[J]. J Comput Sci, 2022, 58: 0-304.
[14]	陈成, 何玉庆, 卜春光, 等. 基于四阶贝塞尔曲线的无人车可行轨迹规划[J]. 自动化学报, 2015, 41(3): 486-496.
	CHEN Cheng, HE Yuqing, BU Chunguang, et al. Feasible trajectory generation for autonomous vehicles based on quartic Bézier curve[J]. A Automo Sinica, 2015, 11(4): 486-496. (in Chinese)
[15]	修国涛, 谢辉, 宋康, 等. 基于驾驶员经验的无人驾驶车辆平行泊车操作模型[J]. 汽车安全与节能学报, 2023, 14(2): 191-201.
	ZHOU Yang, XIE Hui, SONG Kang, et al. Parallel parking operation model for driverless vehicles based on driver experiences[J]. J Autom Safe Energ, 2023, 14(2): 191-201. (in Chinese)

长×宽×高	8.010 m×2.390 m×3.090 m
整车质量，m	13.0 t
轴距，l	4.500 m
前悬，l_f	1.820 m
后悬，l_b	1.690 m
最高车速，v	69.0 km/h
最大方向盘转角，δ_max	565°

长×宽×高	8.010 m×2.390 m×3.090 m
整车质量，m	13.0 t
轴距，l	4.500 m
前悬，l_f	1.820 m
后悬，l_b	1.690 m
最高车速，v	69.0 km/h
最大方向盘转角，δ_max	565°

距离测量值 m	距离真值 m	速度测量值 (km·h^-1)	速度真值 (km·h^-1)
5.49	5.51	9.1	9.2
10.12	10.48	11.0	11.4
15.53	15.59	14.8	15.2
22.30	22.23	16.5	17.7
27.07	27.19	20.5	21.2
32.21	32.28	23.5	23.9
37.86	37.94	25.9	26.3
42.89	42.83	30.5	29.8
47.70	47.81	33.0	33.6

距离测量值 m	距离真值 m	速度测量值 (km·h^-1)	速度真值 (km·h^-1)
5.49	5.51	9.1	9.2
10.12	10.48	11.0	11.4
15.53	15.59	14.8	15.2
22.30	22.23	16.5	17.7
27.07	27.19	20.5	21.2
32.21	32.28	23.5	23.9
37.86	37.94	25.9	26.3
42.89	42.83	30.5	29.8
47.70	47.81	33.0	33.6

避障算法	灵巧性得分 %	舒适性得分 %	安全性得分 %	加权总得分 %	避障耗时t s	加加速度 / (m·s^-3)		碰撞距离 m
避障算法	灵巧性得分 %	舒适性得分 %	安全性得分 %	加权总得分 %	避障耗时t s	纵向 J_v	横向J_I	碰撞距离 m
Bezier	67.58	96.22	74.27	79.51	19.86	4.93	1.18	1.73
Teme-Bezier	74.32	95.71	84.67	85.91	18.85	5.03	1.37	2.16
DP	78.51	97.79	67.06	78.57	18.22	4.42	0.82	1.22
DP+QP	73.32	99.31	68.46	78.69	19.00	3.96	0.43	1.29