User-Assignment Strategy Considering Future Imbalance Impacts for Ride Hailing

The last decade saw the emergence of ride-hailing (RH) services provided by private sector mobility service providers (MSPs). The RH operations usually accumulate customers into batches using fixed time period before assigning vehicles with optimal immediate profit. In the long run, this results in some regions having vehicles oversupply while others having under-supply due to imbalances of trip origins and destinations. For performance improvement, statistical tools are often used which forecast future demand by dividing the operation area into a disjoint set of regions. Using these forecasts, the MSPs periodically reposition idle vehicles to regions with potentially high demand. However, a proactive assignment strategy that systematically reduces the probability of regions becoming imbalanced is rarely discussed. Therefore, we study a balanced regions based assignment method that makes an explicit compromise between short-term batch profit and longterm system imbalance. The method prioritizes assigning trips to vehicles that would lead to decreased long-term system imbalance. The approach is tested in an agent based simulation in Manhattan, using New York City (NYC) taxi data. The results show that the method significantly reduces the longterm supply-demand imbalances, reducing the need of explicit vehicle repositioning.