Real-Time Driver-Request Assignment in Ridesourcing

Abstract

Online on-demand ridesourcing service has played a huge role in transforming urban transportation. A central function in most on-demand ridesourcing platforms is to dynamically assign drivers to rider requests that could balance the request waiting times and the driver pick-up distances. To deal with the online nature of this problem, existing literature either divides the time horizon into short windows and applies a static offline assignment algorithm within each window or assumes a fully online setting that makes decisions for each request immediately upon its arrival. In this paper, we propose a more realistic model for the driver-request assignment that bridges the above two settings together. Our model allows the requests to wait after their arrival but assumes that they may leave at any time following a quitting function. Under this model, we design an efficient algorithm for assigning available drivers to requests in real-time. Our algorithm is able to incorporate future estimated driver arrivals into consideration and make strategic waiting and matching decisions that could balance the waiting time and pick-up distance of the assignment. We prove that our algorithm is optimal ex-ante in the single-request setting, and demonstrate its effectiveness in the general multi-request setting through experiments on both synthetic and real-world datasets.

SKATE : Successive Rank-based Task Assignment for Proactive Online Planning

“…For time management, both immediate assignment ( (Hikima et al 2022), (Sumita et al 2022)) and assignment in time intervals ((Alonso-Mora et al 2017), (Lesmana, Zhang, and Bei 2019)) are common. Some works combine the two aspects such as (Wang and Bei 2022), where the goal is to find a balance between wanting more agents to be available while avoiding the withdrawal of the requests when the waiting time becomes too long.…”

Fully Online Matching with Stochastic Arrivals and Departures

“…Another stream of literature models the delay in matching by incorporating the delay cost in the total cost function and makes the matching decision to minimize cost (Ashlagi et al 2016;Emek, Kutten, and Wattenhofer 2016;Azar and Fanani 2020;Azar, Chiplunkar, and Kaplan 2017;Wang and Bei 2022). It is in contrast to the modeling perspective in fully online matching literature, where a hard constraint for the match to be restricted in a time interval is imposed.…”

Online Proactive Multi-Task Assignment with Resource Availability Anticipation

“…Moreover, to treat the problem online, time has to either be divided in small intervals or requests must be assigned immediately at their arrival. Both have their benefits and drawbacks, because as presented in [26], either the new allocation is sub-optimal (regarding a long-term horizon), due to a lack of available agents at a given moment compared to the fleet, or that, in order to find a better solution, we need to wait for more agents to become available, which can be long. In this work, we have decided to divide the time into intervals and assign the requests that have arrived within the interval to agents that may be available within a given receding-horizon.…”

“…Using only the instantly available agents at the current time may provide sub-optimal solutions in terms of distance and the overall amount of requests allocated [26]. Note if an agent becomes available right after the time window, it will have to wait almost the time window duration before being assigned to new requests.…”