Hierarchical Control for Vehicle Repositioning in Autonomous Mobility on Demand Systems

TL;DR

This paper tackles the challenge of balancing passenger demand with vehicle availability in autonomous Mobility-on-Demand (AMoD) systems by proposing a hierarchical control framework that couples macroscopic and microscopic decision making. The upper layer employs a data-enabled predictive control (DeePC) using Hankel data representations to guide inter-regional vehicle transfers across $R$ regions without explicit system modeling, while the lower layer uses a distributed coverage-control approach on a graph to steer idle vehicles toward high-demand intersections within each region. The two layers interact by exchanging aggregate region-level commands and local vehicle state, and the approach is validated on Shenzhen’s road network, showing improvements in answer rate, waiting time, and vehicle utilization over single-layer or no-control baselines. The study also analyzes the impact of the trade-off parameter $\alpha$, robustness to demand noise, and provides insights into hyperparameter choices and region partitioning, highlighting practical potential for scalable, data-driven AMoD management. Overall, the work demonstrates that integrating data-driven regional rebalancing with microscopic coverage-based guidance yields more efficient, sustainable mobility by reducing empty kilometers and improving service levels.

Abstract

Balancing passenger demand and vehicle availability is crucial for ensuring the sustainability and effectiveness of urban transportation systems. To address this challenge, we propose a novel hierarchical strategy for the efficient distribution of empty vehicles in urban areas. The proposed approach employs a data-enabled predictive control algorithm to develop a high-level controller, which guides the inter-regional allocation of idle vehicles. This algorithm utilizes historical data on passenger demand and vehicle supply in each region to construct a non-parametric representation of the system, enabling it to determine the optimal number of vehicles to be repositioned or retained in their current regions without modeling the system. At the low level, a coverage control-based controller is designed to provide inter-regional position guidance, determining the desired road intersection each vehicle should target. With the objective of optimizing area coverage, it aligns the vehicle distribution with the demand across different districts within a single region. The effectiveness of the proposed method is validated through simulation experiments on the real road network of Shenzhen, China. The integration of the two layers provides better performance compared to applying either layer in isolation, demonstrating its potential to reduce passenger waiting time and answer more requests, thus promoting the development of more efficient and sustainable transportation systems.

Figures (10)

• Figure 1: Hierarchical control structure for vehicle rebalancing in AMoD system. The inter-regional vehicle transfers, determined by the high-level controller, are represented by black arrows. On the right, these transfers are depicted as black taxis, which are directed to relocate to their desired regions accordingly. The orange and green trajectories on the right indicate the intra-regional position guidance for idle vehicles, as given by the low-level controller.
• Figure 2: A schematic diagram of vehicle rebalancing. A five-region example case is shown and a centralized controller gives region-level transfer guidance to empty vehicles in each region, e.g., $u_k^{12}$ informs how many vehicles in Region 1 are asked to relocate to Region 2 at time $k$.
• Figure 3: Coverage control algorithm has been applied in mobile multi-agent problems.
• Figure 4: Illustration of Voronoi partition on the graph. Here we show a 3-vehicle case. Vehicle $i$ is located at $p_i$ (green round marker), its covered nodes are shown as green-filled squares surrounding it, and are connected by lines to show the 'covered area' of vehicle $i$. The centroid of its corresponding $r$-limited Voronoi Cell is denoted as $c_i$, shown as a green diamond marker. For each vehicle, the shortest path from the current position to its centroid is illustrated by the black trajectory. A heat map is superimposed over the road network representing the demand distribution $\phi$. The color gradient ranges from yellow to red, with yellow signifying lower demand probabilities and shades of red indicating increasingly higher demand levels.
• Figure 5: City partition and trip origin and destination distribution.