An integrated optimisation framework for locating depots in shared autonomous vehicle systems

Xinlian Yu, Jingxu Chen, Pramesh Kumar, Alireza Khani, Haijun Mao

Research output: Contribution to journalArticlepeer-review

Abstract

This study presents an integrated optimisation framework for locating depots in a Shared autonomous vehicle (SAV) system under demand uncertainty. A two-stage stochastic mixed integer programming (MIP) model is formulated to optimise the number and locations of depots in a SAV system, where demand uncertainty is represented by multiple scenarios with occurrence probability. The dynamics of vehicle movements are further considered by forming a trip chain for each AV. An enhanced Benders decomposition-based algorithm with multiple Pareto-optimal cuts via multiple solutions is developed to solve the proposed model. The proposed modelling framework and the solution algorithm are tested using two different sizes of transportation networks. Computational analysis demonstrates that the proposed algorithm can handle large instances within acceptable computational cost, and be more efficient than the MIP solver. Meanwhile, insights regarding the optimal deployment of depots in SAV systems are also delivered under different parametric and demand pattern settings.

Original languageEnglish (US)
JournalTransportmetrica A: Transport Science
DOIs
StateAccepted/In press - 2022

Bibliographical note

Publisher Copyright:
© 2022 Hong Kong Society for Transportation Studies Limited.

Keywords

  • benders decomposition
  • demand uncertainty
  • depot location
  • mixed integer stochastic programming
  • Shared autonomous vehicles (SAV)

Fingerprint

Dive into the research topics of 'An integrated optimisation framework for locating depots in shared autonomous vehicle systems'. Together they form a unique fingerprint.

Cite this