Are We There Yet? Managing a Hypothetical Nationwide Autonomous Taxi Network

Abstract

The reality of fully autonomous vehicles is near, and the enthusiasm for the sharingeconomy and ridesharing is strong. As an alternative to the personal vehicle owner-ship model, this thesis proposes a futuristic nationwide autonomous taxi network tosatisfy personal travel demands. Passengers request a ride, and are promptly assigneda vehicle with other passengers traveling in the same direction. Based on a synthetictravel demand data for the United States, this thesis quantifies the ridesharing potential nationwide.Using a set of quality of service constraints, the naive rideshare routing heuristicproduces a set of vehicle routes that satisfies all travel demands. This thesis analyzesthe opportunities for ridesharing and its effect on operational costs and the environment. Motivated by operational efficiency metrics, this thesis also evaluates differentapproaches to determining the ideal aTaxi fleet mix and vehicle size by adding intelligent constraints to rideshare routing.The results are encouraging. In the simplest case, ridesharing decreases totalvehicle miles traveled 38%, increases average vehicle occupancy from 1 to 1.8, anddecreases the number of passenger vehicles on the road by 68%. With the applicationof more intelligent operational constraints and demand forecasting methods, thesevalues can get even better. This thesis serves as a first look at the potential savingsof implementing a nationwide aTaxi system.