Learning Model Predictive Controllers for Real-Time Ride-Hailing Vehicle Relocation and Pricing Decisions

TL;DR

This paper introduces a machine learning approach to approximate Model Predictive Control for ride-hailing vehicle relocation and pricing, enabling higher fidelity decisions with reduced computational cost, and demonstrates improved service quality in NYC data.

Contribution

It proposes learning an MPC optimization proxy to enable real-time, high-fidelity ride-hailing decisions, overcoming computational limitations of traditional MPC methods.

Findings

Improved service quality on NYC ride-hailing data.

Enables higher spatial-temporal resolution in MPC decisions.

Reduces computational complexity of real-time optimization.

Abstract

Large-scale ride-hailing systems often combine real-time routing at the individual request level with a macroscopic Model Predictive Control (MPC) optimization for dynamic pricing and vehicle relocation. The MPC relies on a demand forecast and optimizes over a longer time horizon to compensate for the myopic nature of the routing optimization. However, the longer horizon increases computational complexity and forces the MPC to operate at coarser spatial-temporal granularity, degrading the quality of its decisions. This paper addresses these computational challenges by learning the MPC optimization. The resulting machine-learning model then serves as the optimization proxy and predicts its optimal solutions. This makes it possible to use the MPC at higher spatial-temporal fidelity, since the optimizations can be solved and learned offline. Experimental results show that the proposed approach improves quality of service on challenging instances from the New York City dataset.