校园末端物流无人化的需求测度与模拟优化研究
Research on Demand Measurement and Simulation Optimization of Unmanned Last Mile Delivery in Campus
梁佳宁
清华大学建筑学院 硕士研究生
黄子沐
清华大学建筑学院 科研助理，硕士
龙 瀛（通信作者）
清华大学建筑学院 长聘教授，博士生导师 ylong@tsinghua.edu.cn
摘要: 末端物流无人化作为一种新兴技术手段，虽然已初步应用于实践中，但其在实施效果和优化策略方面尚缺乏系统性和 定量化的研究。通过量化校园末端物流无人化的实际需求，评估不同自动化水平和情景下的社会效益、经济效益、生态 效益，为校园无人化物流规划提供科学依据。以清华大学校园为典型应用场景，结合YOLOv8和ByteTrack算法开发物流 需求测度模型，实现对监控数据中骑手的自动识别与计数。在此基础上，结合问卷数据，利用Anylogic平台进行仿真模拟， 分析末端物流无人化在多种情景下的影响。结果表明，末端物流无人化在提高配送效率和降低能耗方面具有明显优势， 但存在导致顾客等待时间增加的问题，服务水平仍需提升。通过提高机器人或骑手速度、增加机器人数量和增加配送枢 纽数量等策略，可以提高配送效率，减少顾客等待时间，并减少能源消耗。基于模拟结果，发现校园内需要部署至少40台 机器人以完成配送任务。此外，布局4—5个配送枢纽、每个枢纽15—20个机器人，具有较高的经济性。同时，建议将机器人 及骑手速度控制在13—15 km/h以实现安全与效率的平衡。
Abstract: Unmanned last-mile logistics, as an emerging technological approach, has seen preliminary applications in practice. However, systematic and quantitative research on its implementation effectiveness and optimization strategies remains insufficient. This study aims to quantify the actual demand for unmanned last-mile logistics on campus and evaluate the social, economic, and ecological benefits across different automation levels and scenarios, providing scientific guidance for campus logistics planning. Taking the Tsinghua University campus as a typical application scenario, a logistics demand measurement model is developed by integrating the YOLOv8 and ByteTrack algorithms to automatically identify and count delivery riders from surveillance data. On this basis, combined with questionnaire data, a simulation is conducted using the AnyLogic platform to analyze the impact of unmanned last-mile logistics under various scenarios. The results indicate that unmanned last-mile logistics offers significant advantages in improving delivery efficiency and reducing energy consumption. However, it also leads to increased customer waiting times, and the service level still needs to be improved. Strategies such as increasing robot or rider speed, adding more robots, and expanding delivery hubs can enhance delivery efficiency, reduce customer waiting times, and lower energy consumption. Based on the simulation results, it is found that at least 40 robots are required on campus to complete delivery tasks. Furthermore, the layout of 4-5 delivery hubs, each equipped with 15-20 robots, is found to be highly cost-effective. It is also recommended that robot and rider speeds be controlled between 13-15 km/h to balance safety and efficiency.
关键词：无人物流；校园环境；仿真模拟；深度学习；规划设计
Keyword: unmanned delivery; campus environment; simulation modeling; deep learning; planning and design
中图分类号：TU984
文献标识码: A
张锦，陈义友. 物流“最后一公里”问题研究综
述[J]. 中国流通经济，2015，29（4）：23-32.
ZHANG Jin, CHEN Yiyou. A review of research on logistics "last kilometer"[J]. China Business and Market, 2015, 29(4): 23-32.
ALJOHANI K, THOMPSON R G. An examina-
tion of last mile delivery practices of freight carriers
servicing business receivers in inner-city areas[J]. Sustainability, 2020, 12(7): 2837.
SAVELSBERGH M, VAN WOENSEL T. 50th anniversary invited article—city logistics: chal-
lenges and opportunities[J]. Transportation Science,
2016, 50(2): 579-590.
JAOUA A, AMMAR M B, AWASTHI A. A decision
support system for on-demand goods delivery using shared autonomous electric vehicles[J]. International Journal of Decision Support System Technology (IJDSST), 2019, 11(2): 72-88.
PERBOLI G, ROSANO M. Parcel delivery in
urban areas: opportunities and threats for the mix of traditional and green business models[J].
Transportation Research Part C: Emerging Technologies, 2019, 99: 19-36.
DE LA TORRE R, CORLU C G, FAULIN J, et al.
Simulation, optimization, and machine learning
in sustainable transportation systems: models and applications[J]. Sustainability, 2021, 13(3): 1-21.
LAVAEI A, ATASHGAH M A. Optimal 3D trajectory generation in delivering missions under
urban constraints for a flying robot[J]. Intelligent Service Robotics, 2017, 10(3): 241-256.
NGUYEN D H, DE LEEUW S, DULLAERT W, et al. What is the right delivery option for you? Consumer preferences for delivery attributes in
online retailing[J]. Journal of Business Logistics,
2019, 40(4): 299-321.
刘凯. 网约配送员危险骑行行为特性及干预方
法研究[D]. 合肥：合肥工业大学，2021.
LIU Kai. Research on the characteristics of risky riding behavior and intervention methods of net delivery workers[D]. Hefei: Hefei University of Technology, 2021.
FIGLIOZZI M, JENNINGS D. Autonomous delivery robots and their potential impacts on urban
freight energy consumption and emissions[J]. Transportation Research Procedia, 2020, 46: 21-28.
BRUNO D R, DE ASSIS M H, OSÓRIO F S. Development of a mobile robot: robotic guide dog for aid of visual disabilities in urban environ-
ments[C]//2019 Latin American Robotics Sympo-
sium (LARS), 2019 Brazilian Symposium on Robotics (SBR) and 2019 Workshop on Robotics in
Education (WRE). New York: IEEE, 2019: 104-108.
吴春发，鲁冰. 高校校园交通的问题与优化策
略——基于7省区20所高校的调查[J]. 重庆交通
大学学报（社会科学版），2016，16（1）：38-42.
WU Chunfa, LU Bing. Problems and optimization
strategies of campus transportation in colleges and
universities - a survey based on 20 colleges and universities in 7 provinces and regions[J]. Journal
of Chongqing Jiaotong University (Social Science
Edition), 2016, 16(1): 38-42.
SWANSON D. A simulation-based process model
for managing drone deployment to minimize total
delivery time[J]. IEEE Engineering Management Review, 2019, 47(3): 154-167.
SAMOUH F, GLUZA V, DJAVADIAN S, et al. Multimodal autonomous last-mile delivery system design and application[C]//2020 IEEE
International Smart Cities Conference (ISC2). New York: IEEE, 2020: 1-7.
ROMANO A A, SAKAI T, OH S, et al. A simulation-based evaluation of a cargo-hitching service for e-commerce using mobility-on-demand
vehicles[J]. Future Transportation, 2021, 1(3): 639-656.
SHE R, OUYANG Y. Efficiency of UAV-based last-mile delivery under congestion in low-altitude
air[J]. Transportation Research Part C: Emerging Technologies, 2021, 122: 102878.
KHALID R, CHANKOV S M. Drone delivery using public transport: an agent-based modelling and simulation approach[C]//Dynamics in logistics: proceedings of the 7th International Conference LDIC 2020, Bremen, Germany. Cham: Springer International Publishing, 2020: 374-383.
DEVARI A, NIKOLAEV A G, HE Q. Crowdsourcing the last mile delivery of online
orders by exploiting the social networks of retail
store customers[J]. Transportation Research Part E: Logistics and Transportation Review, 2017, 105: 105-122.
MORGANTI E, DABLANC L, FORTIN F. Final deliveries for online shopping: the deployment of pickup point networks in urban and suburban areas[J]. Research in Transportation Business & Management, 2014, 11: 23-31.
MELKONYAN A, GRUCHMANN T, LOHMAR F, et al. Sustainability assessment of last-mile logistics
and distribution strategies: the case of local food networks[J]. International Journal of Production
Economics, 2020, 228: 1-17.
BONABEAU E. Agent-based modeling: methods and techniques for simulating human systems[J].
Proceedings of the National Academy of Sciences,
2002, 99(s3): 7280-7287.
POETING M, SCHAUDT S, CLAUSEN U. A comprehensive case study in last-mile delivery
concepts for parcel robots[C]//Proceedings of the Winter Simulation Conference. 2019: 1779-1788.
DA VIMERCATE A D C. Autonomous robots in food delivery: a simulation study[D]. Milan: Politecnico di Milano, 2018.
CORNO M, SAVARESI S. Measuring urban sidewalk practicability: a sidewalk robot feasibility index[J]. IFAC-PapersOnLine, 2020, 53(2): 15053-
15058.
DESAI P, LOKE S W, DESAI A, et al. Multi-agent based vehicular congestion management[C]//2011 IEEE Intelligent Vehicles Symposium (IV). New York: IEEE, 2011: 1031-1036.
YU J, ZHANG C, WEN J, et al. Integrating multi-agent evacuation simulation and multi-
criteria evaluation for spatial allocation of urban emergency shelters[J]. International Journal of Geographical Information Science, 2018, 32(9): 1884-1910.
LONG G, BINGLEI X, CHUANLING Z, et al. Modeling and simulation analysis for urban rail
transit hub[J]. Journal of Civil Engineering and
Urban Planning, 2023, 5(1): 31-42.
CHEN P, CHANKOV S M. Crowdsourced delivery for last-mile distribution: an agent-based modelling and simulation approach[C]//2017 IEEE
International Conference on Industrial Engineering
and Engineering Management (IEEM). New York:
IEEE, 2017: 1271-1275.
