Recent advances in technology for automating some vehicle functions such as collision avoidance, self-parking and adaptive cruise control have brought not only convenience to drivers but, more importantly, an increase in operational efficiencies and safety. The pace of development and deployment of vehicle automation technologies has gained in recent years. National Highway Traffic Safety Administration categorizes vehicle automation in five levels from Level 0 as no automation to Level 4 as full self-driving automation. Advancement towards Level 4 automation seems inevitable, but the impact of autonomous vehicles (AVs) on future transportation infrastructure and travel patterns is still subject to wide-ranging discussion and speculation.
This paper presents the potential implications of AVs on travel in metro Atlanta using the new activity-based model (ABM) at the Atlanta Regional Commission. The new ABM was recalibrated and revalidated using data from a 2010 transit on-board survey, a regional household travel survey conducted in 2011, 2010 Census data, traffic count and speed data.
This research is conducted in the following scenario modeling context: (1) improved fuel efficiency, (2) change in perceived value of time, (3) increased roadway capacity, (4) change in vehicle ownership, and (5) reduction in parking cost. Reduced fuel consumption is analyzed in terms of reducing the auto operating cost in mode choice and general accessibility calculations as well as the generalized cost in highway assignment. The change in value of time is applied to the model by reducing the in-vehicle time coefficients in tour and trip mode choice utilities for households assumed to have access to an AV. A range of roadway capacities are tested by assuming varying levels of market penetration and availability of vehicle-to-vehicle communications. The change in vehicle ownership is addressed not only for personal vehicles but also for shared vehicles such as demand response services. The on-demand AVs would be especially an attractive alternative to zero-car households. Parking costs at primary destinations are adjusted in mode choice utilities to reflect the reduced costs associated with more efficient parking. The effects of varying the assumed level of market penetration are tested by attributing specific synthetic households with AV availability.