Rebalancing Self-Interested Drivers in Ride-Sharing Networks to Improve Customer Wait-Time

In this article, we address the problem of controlling self-interested drivers in ride-sharing applications. The objective of the ride-sharing company is to improve the customer experience by minimizing the wait-time before pickup. Meanwhile, the drivers attempt to maximize their profit by choosing the best location to wait in the environment between the ride requests assigned to them. The objectives of the ride-sharing company and the drivers are not aligned, and the company has no direct control over the waiting locations of the drivers. The focus of this article is to provide two indirect control methods for the ride-sharing company to optimize the set of waiting locations of the drivers, thereby minimizing one of two objectives: 1) the expected wait-time of the customers or 2) the maximum wait-time of customers. The proposed indirect control methods are: 1) sharing information to a subset of the drivers on the location of other waiting drivers and 2) paying drivers to relocate. We show that the problem of finding the optimal control is NP-hard for both objectives and both control methods. For the information sharing method, we provide an LP-rounding algorithm to minimize the expected wait-time and a three-approximation algorithm to minimize the maximum wait-time. To incentivize the drivers to relocate with payments, we provide three-approximation algorithms for both objectives. Finally, we evaluate the proposed control methods on real-world data and show that we can achieve up to 80% improvement for both objectives.