Stochastic Model Predictive Control for Autonomous Mobility on Demand

Abstract

This paper presents a stochastic, model predictive control (MPC) algorithm that leverages short-term probabilistic forecasts for dispatching and rebalancing Autonomous Mobility-on-Demand systems (AMoD), i.e. fleets of self-driving vehicles. We first present the core stochastic optimization problem in terms of a time-expanded network flow model. Then, to ameliorate its tractability, we present two key relaxations. First, we replace the original stochastic problem with a Sample Average Approximation, and provide its performance guarantees. Second, we divide the controller into two submodules. The first submodule assigns vehicles to existing customers and the second redistributes vacant vehicles throughout the city. This enables the problem to be solved as two totally unimodular linear programs, allowing the controller to scale to large problem sizes. Finally, we test the proposed algorithm in two scenarios based on real data and show that it outperforms prior state-of-the-art algorithms. In particular, in a simulation using customer data from the ridesharing company DiDi Chuxing, the algorithm presented here exhibits a 62.3 percent reduction in customer waiting time compared to state of the art nonstochastic algorithms.

Analysis and Control of Autonomous Mobility-on-Demand Systems

“…In most cases, real-time rebalancing is studied in a model predictive control (MPC) framework, sometimes combined with stochastic models for future demand, possibly enhanced by machine learning (120,121). Also here, previous works exploited totally unimodularity to reduce ILPs to LPs (122).…”

Development Of An Adaptive Weighted Prediction-Based Model Predictive Control Using Disturbance Estimation For Longitudinal Autonomous Driving

“…Onkol [4] proposed the adaptive MPC algorithm to control the fast-varying error state in the inner loop for a two-wheeled robot manipulator with varying mass. M. Tsao et al [5] presented a stochastic MPC algorithm that can leverage short-term probabilistic forecasts for dispatching and rebalancing autonomous mobility-on-demand systems. To design this control algorithm, the authors presented the core stochastic optimization problem in terms of a time-expanded network flow model.…”

Planning and Operations of Mixed Fleets in Mobility-on-Demand Systems

“…Third, unlike human drivers who can be ill-informed or self-interested, AVs can be coordinated in a centralized manner to provide better services which enable higher vehicle utilization and operational efficiency. Thanks to these advantages, AMoD has attracted increasing attention in the research fields of transportation and robotics, including demand analysis and prediction (Chen et al, 2015;Wen et al, 2019), real-time coordination algorithms (Zhang and Pavone, 2016;Liu et al, 2018;Tsao et al, 2018;Iglesias et al, 2019), interactions with public transport (Salazar et al, 2018) and power networks (Tucker et al, 2019;Estandia et al, 2019;Boewing et al, 2020), transportation network design (Pinto et al, 2019;Zardini et al, 2020), and system evaluation (Dia and Javanshour, 2017;Hörl et al, 2018;Fagnant and Kockelman, 2018;Hörl et al, 2019).…”