预约型自动驾驶接驳公交时刻表与车速联合优化

刘春生, 刘涛

清华大学学报(自然科学版) ›› 2026, Vol. 66 ›› Issue (11) : 2220-2233.

PDF(5240 KB)
PDF(5240 KB)
清华大学学报(自然科学版) ›› 2026, Vol. 66 ›› Issue (11) : 2220-2233. DOI: 10.16511/j.cnki.qhdxxb.2026.26.041
需求响应定制公交

预约型自动驾驶接驳公交时刻表与车速联合优化

  • 刘春生1, 刘涛1,2
作者信息 +

Joint optimization of timetable and vehicle speed for reservation-based autonomous feeder transit

  • LIU Chunsheng1, LIU Tao1,2
Author information +
文章历史 +

摘要

自动驾驶接驳公交有助于解决干线公共交通“首末一公里”出行难题。目前,关于自动驾驶接驳公交的研究大多针对接驳单一模式干线公共交通,缺少对于接驳多模式公共交通的研究,尤其是对于如何实现自动驾驶接驳公交与多模式公共交通网络的最优融合,以构建便捷的多模式公共交通一体化无缝出行服务系统的研究。针对上述问题,该文首先基于乘客预约出行需求,构建了预约型自动驾驶接驳公交时刻表与车速联合优化模型,优化自动驾驶接驳公交车队规模和协同换乘成功人数,以提升多模式公共交通的协同服务水平;其次,针对模型结构特征,采用ε约束法进行求解,并设计了模型加速求解程序;最后,以宜宾市多模式公共交通网络为例进行建模优化分析,并通过灵敏度分析揭示了协同换乘等待时间窗长度、规划周期时长、干线公共交通早/晚点车次占总车次的比例等参数对协同换乘成功人数的影响。结果表明:该文所提变量缩减程序可有效降低模型计算的复杂度;与顺序优化法相比,ε约束法可获得Pareto前沿;基于最小车队规模增加一辆车对提高协同换乘成功人数的边际收益最佳;在相同车队规模下,协同换乘等待时间窗长度和规划周期时长均与协同换乘成功人数呈正相关;干线公共交通早/晚点车次占总车次的比例超过50%时,协同换乘成功人数较少且趋于稳定。该文研究结果可为预约型自动驾驶接驳公交与干线公共交通的协同运营提供参考。

Abstract

[Objective] Poorly connected transit services often lead to long walking distances, excessive wait times, and missed transfer connections for passengers transferring between different transport services. These issues significantly reduce the attractiveness and efficiency of public transport systems. Reservation-based feeder transit (RFT) offers a promising solution by providing flexible, on-demand services.However, existing research has mainly focused on its integration with unimodal trunk lines (e. g., metro or bus rapid transit). This study develops a joint optimization framework for optimizing reservation-based autonomous feeder transit (RAFT) timetables and vehicle speeds, aiming to maximize successfully coordinated transfers between RAFT and multimodal trunk transit modes while minimizing fleet size. This work is motivated by RAFT’s potential to substantially enhance multimodal transit service integration, reduce passenger transfer inconvenience, and promote seamless mobility. [Methods] This study formulates a bi-objective optimization model. The primary objective is to maximize the number of successfully coordinated passenger transfers, including transfers from RAFT to trunk transit lines and from trunk transit lines to RAFT. The secondary objective is to minimize the required RAFT fleet size, which directly impacts operational costs. The model’s decision variables include departure times for each scheduled RAFT trip and the average vehicle running speed between transfer stations. To solve this bi-objective mixed-integer linear programming (MILP) model, the ε-constraint method is employed. This method transforms the problem into a single-objective model by treating the fleet size objective as a constraint, generating a set of Pareto-optimal solutions. To address computational complexity, two constraint reduction procedures are developed. The first eliminates binary variables and constraints related to coordinated transfers when no passenger books a transfer or when the scheduled times make a transfer impossible. The second removes vehicle scheduling variables and related constraints if a RAFT trip has no passenger bookings. [Results] A case study was conducted based on a multimodal public transport network in Yibin City, to validate the model and solution method. Four RAFT lines were integrated into a multimodal transit network, including conventional buses, trackless trams, and high-speed rail. Passenger booking data for the morning peak period (7:00-10:00) were utilized. The constraint reduction procedures were proven to be highly effective, reducing the number of transfer-related binary variables by 91.93% and vehicle scheduling variables by 50.74%, thus improving solution efficiency. Using the ε-constraint method, five nondominated Pareto-optimal solutions were obtained, corresponding to fleet sizes ranging from 3 to 7 vehicles. The results showed a trade-off: as the fleet size increased from 3 to 7 vehicles, the total number of successfully coordinated transfers increased from 334 to 554. The marginal gain was most substantial when the fleet size increased by one vehicle from the minimum fleet size. The optimal timetable and speed profiles for each trip were derived. For a fleet size of 7 vehicles, the model achieved the maximum possible 554 successfully coordinated transfers, identical to the result from a sequential optimization approach. Sensitivity analyses revealed further insights: (1) Larger transfer waiting time windows consistently increased successful transfers, with the effect being more pronounced for smaller fleets. (2) Extending the planning horizon significantly increased the problem size and computational time, especially for small fleet sizes. (3) Early or late arrivals of trunk transit service trips negatively affected the number of successful transfers. [Conclusions] This study develops a joint optimization model for optimizing RAFT timetable and vehicle running speed in a multimodal transit network. The model effectively balances service quality (maximizing the number of coordinated transfers) with operational efficiency (minimizing the fleet size). The proposed ε-constraint method, coupled with tailored constraint reduction procedures, provides a practical tool for generating Pareto-optimal solutions and supporting operational decision-making. A key managerial insight is that investing in a fleet slightly larger than the minimum required can yield significant gains in coordinated passenger transfers. Additionally, carefully setting transfer time windows is crucial for optimizing transfer connections. Future work will incorporate real-world considerations such as vehicle capacity constraints, dynamic traffic information, heterogeneous passenger transfer behavior, and uncertainties, potentially employing heuristic or deep reinforcement learning algorithms to solve large-scale real-world problems.

关键词

公共交通 / 自动驾驶 / 接驳公交 / 双目标优化 / ε约束法

Key words

public transport / autonomous driving / feeder transit / bi-objective optimization / ε-constraint method

引用本文

导出引用
刘春生, 刘涛. 预约型自动驾驶接驳公交时刻表与车速联合优化[J]. 清华大学学报(自然科学版). 2026, 66(11): 2220-2233 https://doi.org/10.16511/j.cnki.qhdxxb.2026.26.041
LIU Chunsheng, LIU Tao. Joint optimization of timetable and vehicle speed for reservation-based autonomous feeder transit[J]. Journal of Tsinghua University(Science and Technology). 2026, 66(11): 2220-2233 https://doi.org/10.16511/j.cnki.qhdxxb.2026.26.041
中图分类号: U116.1   

参考文献

[1] 何洪文, 孙逢春, 李梦林. 我国综合交通工程科技现状及未来发展[J]. 中国工程科学, 2023, 25(6): 202-211. HE H W, SUN F C, LI M L. Current status and future development of integrated transportation technology in China [J]. Strategic Study of CAE, 2023, 25(6): 202-211. (in Chinese)
[2] 徐猛, 刘涛, 钟绍鹏, 等. 城市智慧公交研究综述与展望[J]. 交通运输系统工程与信息, 2022, 22(2): 91-108. XU M, LIU T, ZHONG S P, et al. Urban smart public transport studies: A review and prospect[J]. Journal of Transportation Systems Engineering and Information Technology, 2022, 22(2): 91-108. (in Chinese)
[3] WU M, YU C H, MA W J, et al. Joint optimization of timetabling, vehicle scheduling, and ride-matching in a flexible multi-type shuttle bus system[J]. Transportation Research Part C: Emerging Technologies, 2022, 139: 103657.
[4] ZHU C, WU J J, GUO X, et al. Joint optimization of bus scheduling and seat allocation for reservation-based travel[J]. Transportation Research Part C: Emerging Technologies, 2024, 163: 104631.
[5] 奇格奇, 邹恺杰, 邹婕, 等. 面向异质化需求的无人驾驶电动公交接驳路径优化[J]. 清华大学学报(自然科学版), 2022, 62(7): 1178-1185. QI G Q, ZOU K J, ZOU J, et al. Feeder transit routing optimization of driverless electric buses for heterogeneous demands[J]. Journal of Tsinghua University (Science and Technology), 2022, 62(7): 1178-1185. (in Chinese)
[6] GUO R G, GUAN W, VALLATI M, et al. Modular autonomous electric vehicle scheduling for customized on-demand bus services[J]. IEEE Transactions on Intelligent Transportation Systems, 2023, 24(9): 10055-10066.
[7] HE P, JIN J G, SCHULTE F. The flexible airport bus and last-mile ride-sharing problem: Math-heuristic and metaheuristic approaches[J]. Transportation Research Part E: Logistics and Transportation Review, 2024, 184: 103489.
[8] 杨敏, 陈单涛, 蒋瑞宇, 等. MaaS背景下面向枢纽接驳的灵活型公交调度优化[J]. 华南理工大学学报(自然科学版), 2023, 51(10): 22-30. YANG M, CHEN S T, JIANG R Y, et al. Flexible bus scheduling optimization for integrated hub connections in the context of MaaS[J]. Journal of South China University of Technology (Natural Science Edition), 2023, 51(10): 22-30. (in Chinese)
[9] 王正武, 刘杰, 邹文竹. 多车场响应型接驳公交运行线路与调度的协调研究[J]. 铁道科学与工程学报, 2020, 17(7): 1849-1856. WANG Z W, LIU J, ZOU W Z. Coordination research of running lines and vehicle scheduling for multi-depots responsive feeder transit[J]. Journal of Railway Science and Engineering, 2020, 17(7): 1849-1856. (in Chinese)
[10] RAGHUNATHAN A U, BERGMAN D, HOOKER J N, et al. Seamless multimodal transportation scheduling[J]. INFORMS Journal on Computing, 2024, 36(2): 336-358.
[11] HE D D, GUAN W, LIU T, et al. Coordinated operation of fixed-route and demand-responsive feeder transit services in a travel corridor[J]. Journal of Transportation Engineering, Part A: Systems, 2023, 149(2): 04022142.
[12] ZHAO J, SUN S C, CATS O. Joint optimisation of regular and demand-responsive transit services[J]. Transportmetrica A: Transport Science, 2023, 19(2): 1987580.
[13] GKIOTSALITIS K. Coordinating feeder and collector public transit lines for efficient MaaS services[J]. EURO Journal on Transportation and Logistics, 2022, 11: 100057.
[14] 安云龙. 城市道路自动驾驶专用道布设与交通控制优化研究[D]. 北京: 清华大学, 2023. AN Y L. Optimization study on the deployment of dedicated autonomous driving lanes and traffic control on urban roads[D]. Beijing: Tsinghua University, 2023. (in Chinese)
[15] ZHOU L, WANG Y Z, LIU Y D. Active signal priority control method for bus rapid transit based on vehicle infrastructure integration[J]. International Journal of Transportation Science and Technology, 2017, 6(2): 99-109.
[16] MOHAJER N, NAHAVANDI S, ABDI H, et al. Enhancing passenger comfort in autonomous vehicles through vehicle handling analysis and optimization[J]. IEEE Intelligent Transportation Systems Magazine, 2021, 13(3): 156-173.
[17] WANG Y Z, LI D W, CAO Z C. Integrated timetable synchronization optimization with capacity constraint under time-dependent demand for a rail transit network[J]. Computers & Industrial Engineering, 2020, 142: 106374.
[18] GUO Z, WILSON N H M. Assessing the cost of transfer inconvenience in public transport systems: A case study of the London Underground[J]. Transportation Research Part A: Policy and Practice, 2011, 45(2): 91-104.
[19] LEVIN A. Scheduling and fleet routing models for transportation systems[J]. Transportation Science, 1971, 5(3): 232-255.
[20] LIU T, JI W, GKIOTSALITIS K, et al. Optimizing public transport transfers by integrating timetable coordination and vehicle scheduling[J]. Computers & Industrial Engineering, 2023, 184: 109577.
[21] JIN T, LIU T, GKIOTSALITIS K, et al. Optimising public transport passenger transfer waiting time: Comparing the impacts of alternative objectives[J]. Transportmetrica A: Transport Science, 2025: 1-35.
[22] 李元平. 自动驾驶公交车速度控制策略研究[D]. 兰州: 兰州交通大学, 2024. LI Y P. Research on speed control strategies for autonomous driving buses[D]. Lanzhou: Lanzhou Jiaotong University, 2024. (in Chinese)

基金

国家自然科学基金面上项目(72271206);宜宾市科技计划项目(2024JC002);宜宾市“双城”市校协议专项科研经费项目(YBSCXY2023020009)

PDF(5240 KB)

Accesses

Citation

Detail

段落导航
相关文章

/

〈 〉