Estimating short-term travel demand models that incorporate personally owned autonomous vehicles

Abstract

We estimated travel demand models that incorporate a private autonomous vehicle (AV) option using revealed preference data in which personal chauffeurs simulated a personally owned AV. We investigated four components of activity-based models (ABM): activity pattern and primary destination choice, mode choice, and time of day. We compared the chauffeur week models (“AV future”) to the non-chauffeur week models (current conditions). We found no statistically significant differences in parameters of the individual activity pattern, time of day, or destination choice.For mode choice, however, while the auto constant did not change, the mean value of time decreased 60%.As the destination choice model included the mode choice logsum, this results in longer average tour lengths. Moreover, while the trip-making propensity of individuals did not change significantly,there was a 25% increase in systemwide trips due “AVs” (chauffeurs) being sent on errands. This points to the importance of incorporating zero-occupancy vehicle (ZOV) trips into the ABM framework. Our findings suggest that these can be incorporated via the standard ABM development process by adding as additional model components ZOV home-based tours and ZOV subtours. Relatedly,as inter-regional travel is modeled outside the ABM framework, our results indicated that modifications should be made to account for the increase in inter-regional tours, which were 54% more frequent during chauffeur weeks.While these results are from a relatively small sample of 71 individuals, they are the first such travel demand estimation results available from a field experiment, and further studies can build on our framework.