Modeling Automated Vehicles with a Passive Data Model
Corresponding Author: Gregory Macfarlane, Transport Foundry
Presented By: Gregory Macfarlane, Transport Foundry
According to USDOT Secretary Foxx, autonomous vehicles (AVs) have “enormous potential to save lives, reduce greenhouse gas emissions, and transform mobility for the American people.” The current government and large industry leaders including Apple, Google, Tesla, and Uber are investing in policies and technologies to make automated vehicles a reality. In August 2016, Uber announced that it would make self-driving cars available for hailing in the next few months. Even though AVs seem to be right around the corner, few travel models are built to study the effects of AVs, leaving planners and policy-makers unsure how to prepare. This presentation will share a study of AVs in the Asheville, NC region using a pattern-based demand model built from passive, large-scale data and MATSim. This setup provides a microscopic framework from which to analyze short-term responses to AVs assuming shared fleets, privately owned fleets, or a mix of the two. Although AVs present an array of questions-- notably the interaction of AVs with transit and the changes to trip-making behavior-- this application will focus on measuring the sensitivity of total vehicle miles traveled (VMT) and average commute time to differing assumptions of AV adoption and use.