主管:中国科学院
主办:中国优选法统筹法与经济数学研究会
   中国科学院科技战略咨询研究院

中国管理科学 ›› 2024, Vol. 32 ›› Issue (12): 140-152.doi: 10.16381/j.cnki.issn1003-207x.2021.2452

• • 上一篇    下一篇

基于全局视角的智慧城市无人驾驶管控机制及车辆调度问题研究

李昆鹏(), 韩雪芳   

  1. 华中科技大学管理学院,湖北 武汉 430074
  • 收稿日期:2021-11-27 修回日期:2024-01-16 出版日期:2024-12-25 发布日期:2025-01-02
  • 通讯作者: 李昆鹏 E-mail:likp@mail.hust.edu.cn

Research on Autonomous Driving Control Mechanism and Vehicle Scheduling in Smart City Based on Global Perspective

Kunpeng Li(), Xuefang Han   

  1. School of Management,Huazhong University of Science and Technology,Wuhan 430074,China
  • Received:2021-11-27 Revised:2024-01-16 Online:2024-12-25 Published:2025-01-02
  • Contact: Kunpeng Li E-mail:likp@mail.hust.edu.cn

摘要:

随着政策引导及自动驾驶技术的成熟与普及,无人驾驶成为智慧城市交通的重要发展趋势,将改变城市交通的管控逻辑与运作方式。传统人类驾驶模式与当前无人驾驶技术实质都是基于单车视角的局部决策,如单车自行规划路线、决策避撞等。而智慧城市中基于全局视角对城市交通资源和通行车辆进行全局调度,才能有效提升城市交通效率,也将成为无人驾驶环境下新型交通运作模式。该模式下,车辆由城市交通管控平台集中管控,统一规划行驶路线、决策到达各路口的时间及路口避撞方案。鉴于此,本文首先从全局调度机制、路网环境及避撞规则等方面,介绍智慧城市无人驾驶交通运作模式。并基于该模式,提炼出不规则路网下考虑转弯次数的多等级无人驾驶车辆同时调度折线路径问题,建立以“总运行时间最短+转弯次数最少”为目标的数学模型。同时,为快速获得高质量调度方案,本文引入三种优化策略,设计考虑转弯惩罚和方向因子的改进A*算法。最后,模拟路网环境,设置80种规模算例进行仿真实验。结果表明:相较于传统A*算法,本文算法能显著减少车辆的转弯次数和通行总时间,且能将平均求解时间缩短0.061秒。研究成果可为城市交通管理部门规划及管控未来交通、设计全局调度方案提供决策支持。

关键词: 智慧城市, 无人驾驶, 全局调度, 路径规划, 改进A*算法

Abstract:

As policy guidance and automatic driving technologies mature and popularize, unmanned driving has become an important development trend of smart city traffic, which will lead to a dramatic change in the control logic of urban traffic. Currently, both manual driving and automated driving are essentially on a local decision traffic mode based on the perspective of the individual vehicle, e.g., a single vehicle planning its own route and making decisions about collision avoidance. In future smart cities, global scheduling of urban traffic resources and vehicles based on a global perspective will maximize the efficiency of the smart city, and will also become a new mode of traffic operation in a driverless environment. In this mode, the urban traffic control platform uniformly plans the driving route, decides the time of arrival at each intersection and the collision avoidance scheme. In view of this, firstly, the smart city traffic development blueprint is drawn from the aspects of global scheduling mechanism, road network environment and collision avoidance rules. The mode is then defined as a multi-level unmanned vehicle-simultaneous scheduling broken line routing problem considering the number of turns in irregular road network. To solve the problem, a mathematical model with the goal of “the shortest total running time + the least number of turns” is established. To quickly obtain a high-quality scheduling scheme, three optimization strategies are proposed to improve A* algorithm. Finally, 80 scale instances are set up to simulate the road network environment. 32 small-scale instances are used to assess the accuracy of the model and the efficiency of the algorithm, and 48 large-scale instances are used to compare the performance of the traditional A-star algorithm and the algorithm proposed in this paper. Results show that the algorithm proposed in this paper can reduce the number of turns to 1/2 of the traditional A* algorithm, and shorten the overall time of the vehicles to be scheduled by 0.49 h. Especially in large-scale problems, the reduction of the number of turns and the total time-saving effect are more significant. Compared with the traditional A* algorithm, the modified algorithm can reduce the average solution time by 0.061s. The research results can provide decision support for urban traffic management departments to plan and control future traffic and design global scheduling schemes.

Key words: smart city, unmanned driving, global scheduling, path planning, improved A-star algorithm

中图分类号: