Shared Autonomous Vehicles: Rethinking The Morning Commute

The Austin Model

Using advanced computing, Kockelman and Fagnant modeled hundreds of daylong SAV passenger pick-up and drop-off scenarios within a 10-by-10-mile radius. Austin’s general population and density patterns served as the model’s baseline. Their model had SAVs representing 5 percent of all trips made by a total population of approximately 20,000 people.

The researchers concluded that this group of 20,000 people, who were previously served by roughly the same number of conventional cars, would now be served by only 1,700 SAVs. That means that one SAV would take 11 conventional vehicles off the road, freeing up just as many parking spaces in the process.

Kockelman and Fagnant also estimated an average wait time under 20 seconds — from the time trip-makers arrive at a shared-car station or parking area to the time their cars are available — and found that fewer than 1 percent of SAV members would have to wait more than five minutes. Additionally, each SAV would be responsible for emissions savings, including 34 percent less carbon monoxide emissions, largely due to fewer cold engine starts, which are a major source of emissions.

The researchers did spot disadvantages, however, including the possibility that SAVs would lead to more traffic congestion, even as they are taking conventional vehicles off the road.

“All vehicles that are automated and connected have an opportunity to cut congestion, particularly on freeways, through processes like cooperative adaptive cruise controls,” Fagnant said. “But at lower levels of market penetration, SAVs may actually lead to more congestion, since each vehicle will be traveling unoccupied as it drives from one passenger to the next.”

In the not-so-distant future, Fagnant believes that his and Kockelman’s SAV model could help drivers, urban planners and policymakers make better transportation decisions, and facilitate ride-sharing, which can moderate possible congestion impacts emerging from easier travel.

“I think city planners would be very interested in the parking implications,” Fagnant said. “And cities could potentially operate their own fleets of SAVs, not just for single occupancy but also for ride sharing.”

Fagnant and Kockelman are now applying local land use, network and travel data to reflect Austin-specific travel patterns. Their new model is designed to test the implications of using SAVS to ride share, whereby travelers headed to the same destination or in similar directions can share a ride in a single vehicle.

Vehicle icon in top illustration courtesy of Oliver Guin, through Creative Commons.