Intermodal Network of Autonomous Mobility-on-Demand and Micromobility Systems

TL;DR

This work tackles the problem of quantifying and optimizing the joint operation of autonomous micromobility-on-demand (AMoD) and micromobility within a unified intermodal network. It introduces a three-layer, multicommodity network-flow framework over walking, micromobility, and road networks, with a convexified travel-time function $t_{ij}(X_{ij}^\mathrm{road})$ and decision variables including $x_{ij}^{m}$, $X_{ij}$, $\beta_i^{\mathrm{in/out}}$, and $x_{ij}^{0}$, aiming to minimize the total travel time $\sum_{(i,j)\in \mathcal{A}} t_{ij}\cdot X_{ij}$. The authors apply the model to the Sioux Falls network, using real data and prescribed free-flow speeds $t_{ij}^0$ for each mode, and demonstrate how varying fleet sizes and micromobility rebalancing capacity shift mode shares and reduce travel times. The key contributions are the I-AMoD optimization formulation and the quantitative insights into how intermodal interdependencies and rebalancing dynamics affect performance, offering practical guidance for urban planners on coordinated AMoD-micromobility deployments.

Abstract

This paper studies models for Autonomous Micromobility-on-Demand (AMoD), a paradigm in which a fleet of autonomous vehicles delivers mobility services on demand in conjunction with micromobility systems. Specifically, we introduce a network flow model to encapsulate the interaction between AMoD and micromobility under an intermodal connection scenario. The primary objective is to analyze the system's behavior, optimizing passenger travel time. Following this theoretical development, we apply these models to the transportation networks of Sioux Falls, enabling a quantifiable evaluation of the reciprocal influences between the two transportation modes. We found that increasing the number of vehicles in any of these two modes of transportation also incentivizes users to use the other. Moreover, increasing the rebalancing capacity of the micromobility system will make the AMoD system need less rebalancing.

Figures (4)

• Figure 1: Intermodal network representation: the intermodal network consists of walking, micromobility, and AMoD graphs. The colored dots denote intersections or stops, and the black arrows represent road links or pedestrian pathways. The dotted lines denote geographically close nodes, while the gray curved arrows are the mode-switching arcs connecting them.
• Figure 2: Time-based modal share and average travel time varying the number of AMoD vehicles. Users' modal share and average travel time converge to a steady-state value after $n_\mathrm{R} = 7000$.
• Figure 3: Time-based modal share and average travel time trend changing the micromobility fleet size. The system behavior changes, changing the AMoD to the micromobility fleet size ratio.
• Figure 4: AMoD and micromobility total rebalancing flow change by increasing the micromobility system's rebalancing capacity. Increasing the micromobility system's rebalancing capacity will make the AMoD vehicles rebalance less frequently.