Taxi-ridesharing systems are considered an important means for sustainable urban transport. Previous literature shows that introducing meeting points in ridesharing, where customers are picked up and dropped off, increases its performance. We consider an on-demand taxi-ridesharing system, where the focus lies on the anticipatory assignment of customers to meeting points. We model the problem as a sequential decision process with the objective to maximize the distance saved through sharing. We suggest an anticipatory solution method for the planning of trips which assigns passengers to meeting points. We evaluate the suggested method on instances arising from real-world data and show that it leads to a significant increase in saved distance, and consequently CO2 emissions when compared to a benchmark. We analyze the problem’s and method’s parameters and show that anticipatory methods further leverage the economical and ecological advantages of ridesharing.
A Pre-Print is available on RIS (UPB)