Adaptive Pricing Mechanisms for On-Demand Mobility

We consider on-demand car rental systems for public transportation. In these systems, demands are often unbalanced across different parking stations, necessitating costly manual relocations of vehicles. To address this so-called "deadheading" effect and maximise the operator's revenue, we propose two novel pricing mechanisms. These adaptively adjust the prices between origin and destination stations depending on their current occupancy, probabilistic information about the customers' valuations and estimated relocation costs. In so doing, the mechanisms incentivise drivers to help rebalance the system and place a premium on trips that lead to costly relocations. We evaluate the mechanisms in a series of experiments using real historical data from an existing on-demand mobility system in a French city. We show that our mechanisms achieve an up to 64% increase in revenue for the operator and at the same time up to 36% fewer relocations.