Model Impacts of Connected and Autonomous/Automated Vehicles (CAVs) and Ride-Hailing with an Activity-Based Model (ABM) and Dynamic Traffic Assignment (DTA)

Ben Stabler, Mark Bradley, Dan Morgan, Howard Slavin, Khademul Haque
Tuesday, August 28, 2018
Exploratory Modeling and Simulation

This report describes the experiment on how to model impacts of connected and autonomous/automated vehicles (CAVs) and ride-hailing with an Activity-Based Model (ABM) and Dynamic Traffic Assignment (DTA) in the context of Exploratory Modeling and Analysis (EMA). EMA is a systematic approach to perform sensitivity analyses using models when users cannot assert many of the model inputs with confidence. The example EMA described integrates the DaySim activity-based travel demand model with the TransModeler dynamic traffic simulation model for the Jacksonville, Florida, region. The approach adapts the travel demand model to simulate households’ decisions whether to purchase CAVs instead of conventional vehicles and to simulate travelers’ decisions whether to use CAV-based carsharing and ride-hailing services. The dynamic network model simulates operating characteristics of CAVs—depending on network vehicle mix—and simulates the performance of CAV-only infrastructure under different demand scenarios. The integrated model system simulates dozens of different scenario combinations to demonstrate the potential of exploring possible outcomes and finding critical input assumptions while identifying future policy directions that are likely to be the most robust in the face of “deep uncertainty.”