Scholarlink

<p>Early-stage deployment of shared autonomous vehicles (SAVs) is likely to be constrained by still-maturing autonomous driving technology and safety regulations, making it realistic to operate SAVs from a limited number of depots and to confine automated driving to a subset of dedicated links embedded in mixed traffic. We formulate a joint design problem for unidirectional SAV-only travel sections and a multi-depot SAV fleet, explicitly accounting for flexible return-to-any-depot operations. A space–time mixed-integer linear programming model is proposed that integrates budget-constrained lane activation, depot-level fleet sizing, and daytime operations of SAVs and human-driven vehicles (HDVs); Depot-specific fleet sizes are determined endogenously, and end-of-horizon depot imbalances are penalised as a rebalancing cost that approximates off-service-hours repositioning. Optional strong–connectivity constraints are introduced to eliminate absorbing SAV-only subnetworks and to test stricter lane-design rules. Numerical experiments on the Sioux Falls network show that, under tight lane budgets, unidirectional dedicated sections combined with a central multi-depot layout substantially reduce empty running and weighted travel cost while achieving high SAV utilisation. The main efficiency patterns are robust across rebalancing weights, connectivity requirements and HDV ownership-cost scenarios: moderate rebalancing penalties already drive depot imbalances close to zero, and higher HDV costs primarily amplify SAV adoption and benefits in demand-dense central areas.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Joint Design of Unidirectional Dedicated Lanes and Multi-Depot Shared Autonomous Vehicle Fleets with Flexible Rebalancing

  • 期刊论文

  • 发表日期：

    2026年4月24日

查看全文
  • Yu Li,
  • Wushuang Wang,
  • Maiko Shigeno

摘要

Early-stage deployment of shared autonomous vehicles (SAVs) is likely to be constrained by still-maturing autonomous driving technology and safety regulations, making it realistic to operate SAVs from a limited number of depots and to confine automated driving to a subset of dedicated links embedded in mixed traffic. We formulate a joint design problem for unidirectional SAV-only travel sections and a multi-depot SAV fleet, explicitly accounting for flexible return-to-any-depot operations. A space–time mixed-integer linear programming model is proposed that integrates budget-constrained lane activation, depot-level fleet sizing, and daytime operations of SAVs and human-driven vehicles (HDVs); Depot-specific fleet sizes are determined endogenously, and end-of-horizon depot imbalances are penalised as a rebalancing cost that approximates off-service-hours repositioning. Optional strong–connectivity constraints are introduced to eliminate absorbing SAV-only subnetworks and to test stricter lane-design rules. Numerical experiments on the Sioux Falls network show that, under tight lane budgets, unidirectional dedicated sections combined with a central multi-depot layout substantially reduce empty running and weighted travel cost while achieving high SAV utilisation. The main efficiency patterns are robust across rebalancing weights, connectivity requirements and HDV ownership-cost scenarios: moderate rebalancing penalties already drive depot imbalances close to zero, and higher HDV costs primarily amplify SAV adoption and benefits in demand-dense central areas.