高速多车多驾驶行为的冲突协同决策方法

doi:10.3969/j.issn.1674-8484.2023.05.010

摘要/Abstract

摘要：

为解决高速公路场景下多车多驾驶行为车辆空间位置冲突的问题，提出了一种基于二分图最优匹配的驾驶行为冲突协同决策方法。根据静态交通信息，创建车辆可行候选驾驶行为集，依据车道平均车速、车辆密度、行进空间、即碰时间（TTC）、行驶负担等5种评价指标构建效用函数，对候选驾驶行为定量评价；以车辆实施驾驶行为结束时刻的潜在空间位置为目标集合，驾驶行为评价效用为权值，构建基于车辆集合与目标集合的二分图；以全局总效用值最大为决策目标，采用Kuhn-Munkres（KM）算法求解最优匹配；搭建仿真场景，来验证该方法的有效性。结果表明：该协同决策方法可有效解决多车多驾驶行为冲突，保证车辆行车安全，提高道路上初末时刻车辆2%的效用值和8%的平均车速，增加了通行效率，且驾驶行为决策的准确性相比于遗传算法（GA）、粒子群算法（PSO）分别提高了11%和9%，同时KM算法的实时性远高于GA算法和PSO算法。

关键词: 自动驾驶, 协同决策, 二分图, Kuhn-Munkres（KM）算法, 效用函数

Abstract:

A collaborative decision-making method of driving behavior conflict was proposed based on the optimal matching of dichotomous graph to solve the problem of spatial position conflict of multi-vehicle and multi-driving behavior in the highway scenario. A set of feasible candidates driving behaviors of vehicles was created according to the static traffic information. A utility function was constructed according to five evaluation indicators, including the average lane speed, the vehicle density, the travel space, the time to collision (TTC), and the driving burden, to quantitatively evaluate the candidate driving behaviors. Taking the potential spatial position at the end of the vehicle's driving behavior as the target set, and the driving behavior evaluation utility as the weight, a dichotomous graph based on the vehicle set and the target set was constructed. Taking the maximum global total utility value as the decision-making goal, the Kuhn-Munkres (KM) algorithm was used to solve the optimal matching. A simulation scenario is built to verify the effectiveness of the method. The results show that the collaborative decision-making method effectively solve the conflict of multi-vehicle and multi-driving behavior, ensure vehicle driving safety, enhance the utility value by 2% and the average speed by 8% at the initial and final moments of the road, results in increased the traffic efficiency, and the accuracy of driving behavior decision-making is 11% and 9% higher than that of Genetic Algorithm (GA) and Particle Swarm Optimization (PSO), respectively. At the same time, the real-time performance of KM algorithm is much higher than that of GA algorithm and PSO algorithm.

Key words: autonomous driving, collaborative decision-making, bipartite graph, Kuhn-Munkres (KM) algorithm, utility

中图分类号:

U461

张新锋, 吴琳, 李致远, 柳欢. 高速多车多驾驶行为的冲突协同决策方法[J]. 汽车安全与节能学报, 2023, 14(5): 609-617.

ZHANG Xinfeng, WU Lin, LI Zhiyuan, LIU Huan. Collaborative decision-making method of high-speed multi-vehicle multi-driving behavior confliction[J]. Journal of Automotive Safety and Energy, 2023, 14(5): 609-617.

图/表 16

参考文献 15

[1]	YIN Taiqiao, LI Ying, Fan Jiaohao, et al. A novel gated recurrent unit network based on svm and moth-flame optimization algorithm for behavior decision-making of autonomous vehicles[J]. IEEE Access, 2021, 9: 20410-20422. doi: 10.1109/ACCESS.2021.3054755 URL
[2]	Furda A, Vlacic L. Enabling safe autonomous driving in real-world city traffic using multiple criteria decision making[J]. IEEE Intel Transp Syst Maga, 2011, 3(1): 4-17.
[3]	WANG Pengwei, GAO Song, LI Liang, et al. Research on driving behavior decision making system of autonomous driving vehicle vased on benefit evaluation model[J]. Archives Transp, 2020, 53(1): 21-36. doi: 10.5604/08669546 URL
[4]	杨达, 吕蒙, 戴力源, 等. 车联网环境下自动驾驶车辆车道选择决策模型[J]. 中国公路学报, 2022, 35(4): 243-255. doi: 10.19721/j.cnki.1001-7372.2022.04.020
	YANG Da, LÜ Meng, DAI Liyuan, et al. Decision-making model for the lane selection of automated vehicles in the connected vehicle environment[J]. Chin J Highway Transp, 2022, 35(4): 243-255. (in Chinese)
[5]	Schmidt M, Wissing C, Nattermann T, et al. A probabilistic model for discretionary lane change proposals in highway driving situations[J]. Forschung im Ingenieurwesen, 2021, 85(2): 485-500. doi: 10.1007/s10010-021-00439-0
[6]	Düring M, Pascheka P. Cooperative decentralized decision making for conflict resolution among autonomous agents[C]// 2014 IEEE Int’l Symp Inno Intel Syst Appl (INISTA) Proc, 2014: 154-161.
[7]	Lovellette E, Hexmoor H. Lane and speed allocation mechanism for autonomous vehicle agents on a multi-lane highway[J]. Internet of Things, 2021, 13: 1-19.
[8]	LIU Yingxiang, LI Hao, ZHU Xuefeng. Brainstorming multi-agent reinforcement learning for multi-vehicles games[C]// 2021 Int’l Conf Digi Soc Intell Syst (DSInS), 2021: 357-361.
[9]	Kessler T, Knoll A. Cooperative multi-vehicle behavior coordination for autonomous driving[C]// 2019 IEEE Intel Vehi Symp (IV), 2019: 1953-1960.
[10]	李珣, 曲仕茹, 夏余. 车路协同环境下多车道车辆的协同换道规则[J]. 中国公路学报, 2014, 27(8): 97-104.
	LI Xun, QU Shiru, XIA Yu. Cooperative lane-changing rules on multilane under condition of cooperative vehicle and infrastructure system[J]. Chin J Highway Transp, 2014, 27(8): 97-104. (in Chinese)
[11]	张毅, 姚丹亚, 李力, 等. 智能车路协同系统关键技术与应用[J]. 交通运输系统工程与信息, 2021, 21(5): 40-51.
	ZHANG Yi, YAO Danyao, LI Li, et al. Technologies and applications for intelligent vehicle-infrastructure cooperation systems[J]. J Transp Syst Eng Info Tech, 2021, 21(5): 40-51. (in Chinese)
[12]	Barbera A J, Horst J A, SCHLENOFF C I, et al. Task analysis of autonomous on-road driving[C]// Mobile Robots XVII, 2004, 5609: 61-72.
[13]	LU Xinghao, ZHAO Haiyan, GAO Bingzhao, et al. Decision-making method of autonomous vehicles in urban environments considering traffic laws[J]. IEEE Trans Intel Transp Syst, 2022, 23(11): 21641-21652. doi: 10.1109/TITS.2022.3183229 URL
[14]	彭浩楠, 王伟达, 项昌乐, 等. 自动驾驶汽车双车道换道优化轨迹规划方法[J]. 吉林大学学报(工学版), 2022, 8(2): 1-12.
	PENG Haonan, WANG Weida, IANG Changle, et al. Optimization-based lane changing trajectory planning approach for autonomous vehicles on two-lane road[J]. J Jilin Univ (Eng Tech Ed), 2022, 8(2): 1-12. (in Chinese)
[15]	刘思阳, 蒙涛, 雷家坤, 等. 微纳卫星星座的Kuhn-Munkres匹配部署优化方法[J]. 宇航学报, 2021, 42(7): 895-906.
	LIU Siyang, MENG Tao, LEI Jiakun, et al. Kuhn-Munkres matching deployment optimization method for micro-satellite constellation[J]. J Astronautics, 2021, 42(7): 895-906. (in Chinese)

Lane 1限速	110~120 km/h
Lane 2限速	90~110 km/h
Lane 3限速	60~100 km/h
观测区域长度	200 m
车道宽度	3.5 m
车长×车宽	6.0 m × 1.8 m
速度评价指标权重， α_v	0.40
车辆密度评价指标权重， α_dens	0.15
行进空间评价指标权重，α_spa	0.20
TTC评价指标权重，α_TTC	0.20
行驶负担评价指标权重， α_dirv	0.05
车辆最小安全距离，2l	12 m
驾驶行为持续时间，t_dura	4.0 s
最大行进空间，S_max	200 m

Lane 1限速	110~120 km/h
Lane 2限速	90~110 km/h
Lane 3限速	60~100 km/h
观测区域长度	200 m
车道宽度	3.5 m
车长×车宽	6.0 m × 1.8 m
速度评价指标权重， α_v	0.40
车辆密度评价指标权重， α_dens	0.15
行进空间评价指标权重，α_spa	0.20
TTC评价指标权重，α_TTC	0.20
行驶负担评价指标权重， α_dirv	0.05
车辆最小安全距离，2l	12 m
驾驶行为持续时间，t_dura	4.0 s
最大行进空间，S_max	200 m

驾驶行为	效用值	潜在位置区间/ m	目标车道编号
(m₁, n₁)	0.773	(116.2,134.2)	1
(m₁, n₂)	0.773	(121.8,139.8)	1
(m₃, n₁)	0.705	(116.2,134.2)	2
(m₃, n₃)	0.705	(108.6,126.6)	2

驾驶行为	效用值	潜在位置区间/ m	目标车道编号
(m₁, n₁)	0.773	(116.2,134.2)	1
(m₁, n₂)	0.773	(121.8,139.8)	1
(m₃, n₁)	0.705	(116.2,134.2)	2
(m₃, n₃)	0.705	(108.6,126.6)	2

目标	潜在位置区间/ m	车道编号
T₁	(121.8,139.8)	1
T₂	(116.2,134.2)	2
T₃	(147.3,165.3)	1
T₄	(120.7,138.7)	2
T₅	(107.9,125.9)	3
T₅	(108.4,126.4)	3
T₆	(174.2,192.2)	1
T₇	(166.2,184.2)	2
T₇	(171.3,189.3)	2
T₈	(169.6,187.6)	3