September 16, 2019

Progress in estimating additional travel induced by automated vehicles

Andisheh Ranjbari, Sustainable Transportation Lab alumna Jingya Gao, and I have just published a paper in the journal Transportation that investigates how relieving travelers from the task of driving affects their value of travel time. In past work, we have shown that by reducing the value of travel time, automated vehicles could lead to significantly more vehicle travel. This new paper helps us to refine those estimates, while highlighting some of the challenges of trying to forecast the impacts of vehicle automation, which is still a strange concept to many travelers.

Value of travel time is a measure of the cost of time spent traveling, as perceived by travelers. It is an important predictor of travel demand; in many cases the cost of the driver’s time is the single largest component of the overall (or “generalized”) cost of travel! Reduce the value of travel time and you reduce the overall cost of travel; reduce the overall cost of travel and people will travel more. More travel is not a bad thing per se, but it generally leads to more emissions and more traffic congestion.

The key challenge with measuring how vehicle automation affects the driver’s value of time is that we don’t really have fully operational, driverless vehicles in service today. So for this work, we conducted a survey in which we asked people to make a hypothetical choice for how to make a 15-mile commute trip: either driving themselves, or using a ridehailing service similar to Uber or Lyft. The main idea is that the in-vehicle experience of a ridehailing service today is a lot like we would expect riding in a driverless car to be in the future: in both cases, you input a destination, sit back, and are transported to your destination, without further effort. By randomly varying the travel time and the cost of the different options that we show to people in the survey, we can uncover how sensitive people’s choices are to cost, travel time while driving themselves, and travel time in a ridehailing vehicle.

We are not the first to do something like this, but we introduced a couple of new twists. First, half of our survey takers were asked to choose between driving themselves and taking a regular ridehailing service, and the other half chose between driving themselves and taking a driverless ridehailing service. Second, we suspect that many people today may not fully appreciate the opportunities for multitasking when someone else is driving, so we explicitly reminded half our survey takers that using a ridehailing service would allow them to do something other than drive (such as work, read, rest, or use their phone) during the trip, while we gave no such reminder to the other half.

The results were fascinating. Survey responses show that participants perceive the cost of time spent traveling to be about 13% lower when traveling in a ridehailing service than when driving themselves. However, when we reminded people that a ridehailing service would allow them to engage in other activities while traveling, the perceived cost of travel time was 45% lower than when driving themselves. But when we told people that the ridehailing service was driverless, their value of time actually increased. That means that our survey respondents would find each minute spent riding in a driverless vehicle to be more burdensome than a minute driving themselves!

What does all this mean? I suspect that today, many people are skeptical of driverless cars and would not trust the technology, which is why the value of travel time is higher in these vehicles. But in the long run, if driverless vehicles prove to be commercially successful, this bias would disappear and the value of time in a driverless car will turn out to be similar to that in a conventional ridehailing vehicle today: somewhere in the range of a 13-45% reduction below the value of time when driving oneself. But in the near term, we should be very careful about asking survey questions about driverless cars; people’s skepticism of the technology in the near term may lead to underestimation of the potential for the technology to increase travel demand in the long term.

A PDF of the paper (non-paywalled) is available here.