A filtering system to solve the large-scale shared autonomous vehicles Dial-a-Ride Problem

Abstract

Shared autonomous vehicles are rapidly becoming one of the most popular topics in the fields of transportation and operations research. This paper studies a prospective transportation system in which shared autonomous vehicles (SAVs) providing Dial-a-Ride services in urban and rural areas must meet a large number of passenger requests (at least 105 per day). We consider those SAVs likely to replace, in the future, most of the individual vehicles inside urban areas. We address this very large-scale problem with a greedy insertion heuristic that involves a specific filtering system design in order to speed up the decision process and relies on an original network-driven encoding of the routes. Experimental results show that en-route vehicles can be drastically reduced by such systems using a fleet of SAVs of capacity 10, decreasing the vehicle number by more than 98%, compared to a situation where all travel demands would rely on personal vehicles. In addition, thanks to the filtering system, the total execution time can be reduced by almost 97% compared to the classic best-fit insertion heuristic, while maintaining the quality of the solution. This opens the way to the adaptation of our algorithms to a dynamic real-time context.