Managing Advanced and Flexible Ride-Hailing Requests

Ride-hailing services are one of the cornerstone building blocks of modern smart mobility. A central challenge in such services is the scheduling of rides. Classically, two versions of the problem have been explored: a static, when all ride requests are known in advance, and a dynamic, when ride requests arrive one by one. We explore a dynamic setting, but still assume that many users can plan their travel well in advance, increasing the potential for fleet-wide optimization. Knowledge of future trips in advance is one aspect of flexibility that the system leverages. The other aspect stems from the fact that often there is no need to travel at a specific time, rather a user may want to transport persons or goods to a given destination by some deadline and leave not earlier than a given threshold. The paper explores the engineering of a system that builds on an off-the-shelf optimization engine (OR-tools) and employs batching of requests while managing the adjustable scheduling and early commitment of advanced requests, balancing the user-oriented early commits and the efficiency of the fleet. Simulation experiments demonstrate the merits of the proposed solution.