Dynamic pricing for ride-hailing services considering relocation and mode choice

Ride hailing services are gaining popularity in cities around the world, and the introduction of shared autonomous vehicles is likely to pronounce this trend. A problem with these services is the extra travel to relocate without customers, necessary to serve an asymmetric demand. One solution to reduce the amount of empty driving is the adoption of an optimized pricing strategy that takes into account the cost of relocation. In this work, we present a dynamic pricing optimization that is integrated with online predictive relocation. We simulate the choice between a service that is operated between densely distributed demand points with shared autonomous vehicles and a generic alternative mode. The strategy finds optimal fares for each origin/destination area in each period of the day to maximize net profits for the operator. Our approach is flexible, fast, and scalable. We demonstrate our proposed methodology with a large dataset of taxi trips from New York City. The results of the simulations show that our pricing strategy consistently increases the operator’s revenues and decreases relocation costs when compared to the case with an optimal but constant, distance dependent fare. In cases where passengers’ waiting times are significant, our approach also significantly reduces average wait times, thus benefiting passengers too.