A Future for American Mobility: A Case for Shared Autonomous Vehicle Adoption Through aTaxi Network Optimization

Abstract

The era of autonomous vehicles is almost upon us. With each passing day we hear more and more about the accomplishments achieved by the range of developers in the autonomous vehicle space. Technology giants like Alphabet and Uber are positioned in competition against original equipment manufacturers in a race to develop the first commercially available driverless cars. The societal benefits of this technology have led some to predict the steady decline of the conventional car and the rise of mobility based on autonomous vehicles. Given that this shift occurs, the popu- larization of eco-friendly ridesharing is a real possibility for the first time. Shared autonomous taxis (aTaxis) present attractive alternatives for mass transportation because of the safety and convenience that they offer, in addition to ridesharing’s ability to mitigate many of the negative effects of transportation. This thesis first makes a case for the adoption of a nationwide aTaxi ridesharing service to replace the conventional personal-car model. Tens of thousands of lives and millions of vehicle miles could be saved every year if an aTaxi service provided national mobility. For our case, we outline specific benefits and work to demonstrate the need for an optimized personal transportation model. We develop an aTaxi solution, grounding our proposed model in transit theory. This thesis then aims to define an optimal level-of-service (LoS) at which an aTaxi fleet could meet the entirety of New Jersey trip demand. We complete this goal by first establishing a network of aTaxi stand locations, and connecting them to accurately model the intricate network of NJ road- ways. Using synthesized trip data extrapolated from past census information, we flow each generated trip across the network of aTaxi stands. We complete this flow 120 times, each time varying the level-of-service in order to gather data and determine the optimal LoS. We define optimal fleets for varying levels of ridesharing, representing different stages in aTaxi adoption. The shift from individual trip taking to highly efficient ridesharing will have important implica- tions on total vehicle miles traveled as well as on average vehicle occupancy for the state’s vehicles. Higher AVO values can lead to significant traffic relief, which we aim to quantify in our analysis. We will compare how each level of ridesharing affects congestion in NJ by flowing the correspond- ing aTaxi trip data across the NJ road network. Congestion mitigation through ridesharing is an understudied aspect of the argument for autonomous vehicles. This thesis aims to provide evidence that a significant reduction in traffic is possible given high levels of aTaxi ridesharing.