As tech companies and automakers alike pour billions into autonomous vehicles and technologies, it has been largely assumed that driverless vehicles will alleviate congestion. The idea is that more and more people will abandon private auto ownership in favor of booking pay-as-you-go autonomous rides. Recent research on the impact of driverless vehicles on traffic, however, offers mixed messages.
Traffic Calming or Cursing?
A study from Rutgers University-Camden scholar Benedetto Piccoli found that just one autonomous vehicle that doesn’t abruptly slow down or change lanes (as humans so often do) could improve the traffic flow of 20 surrounding vehicles, and that if five percent of vehicles on a given road were autonomous, fuel consumption could be reduced by 40 percent and braking overall traffic flow could improve by 15 percent. The theory is that fewer “stop and go waves” will improve vehicle throughput for everyone.
But other research to suggest that self-driven vehicles won’t cure traffic by default. A report by the University of California at Santa Cruz’s Adam Millard-Ball found that because “cruising”—driving around empty while a passenger is at work—will be far less expensive than parking, empty vehicles will add to traffic woes in cities either by circling continuously or by returning home to park free. Simulating traffic in downtown San Francisco, Millard-Ball found that autonomous vehicles could double traffic in the city.
A separate study by the World Economic Forum and the Boston Consulting Group looked at traffic in Boston and found that autonomous vehicles would increase traffic by about five percent, mostly from people using these vehicles in place of public transit.
These scenarios are not far-fetched if you look at the impact human-driven ridehailing services have had on city traffic. Ridehailing vehicles added 5.6 billion vehicle miles in the nine largest cities in the United States, with 60 percent coming at the expense of public transit. This is due to the fact that the vehicles travel on average 2.8 miles empty for every paid mile.
Sharing Matters Most
Simply replacing today’s private ridehailing services with autonomous counterparts may incrementally improve traffic flow with safer driving, but city growth is expected to accelerate in coming years, so one-for-one replacement of human-driven vehicles won’t do much to relieve congestion. The vehicles and the services that they provide must be shared, with multi-passenger trips becoming the dominant use of the vehicles.
For shared services to be successful, they must be convenient. A University of Texas study of traffic in the Austin area found that one shared autonomous vehicle could do the work of nine private vehicles, with minimal impact on ride times and with wait times from 20 seconds to five minutes. That’s likely less time than it takes to get to a car, park, and walk to a destination.
Bestmile performed a similar analysis of traffic in the city of Chicago. We used our mobility services platform to crunch publicly available taxi trip data to analyze how a shared service would compare to the performance of a single-rider taxi service. We found that 200 shared vehicles, autonomous or human driven, could do the work of the city’s 2700-vehicle taxi fleet with average wait times of five minutes.
Optimizing Shared Services
These scenarios envision highly optimized sharing or pooling solutions that constantly monitor and manage vehicle locations, vehicle capacity, and vehicle fuel levels along with ride requests, destinations, traffic, and weather to deliver ultra-efficient routing and ride matching. Achieving predictable pickup times and ride times requires constant learning from historical and real-time demand data, as well as understanding details such as what side of a street a passenger is on, street configurations (one-way or two-way, for example), traffic to the rider and to the destination—all to determine which vehicle can deliver the best user experience under the existing circumstances.
Location Location Location
Positioning and repositioning vehicles each day will also be critical to service success. As noted by the studies suggesting autonomous vehicles could worsen traffic, minimizing empty vehicle travel plays an important part in meeting passenger and operator performance requirements. Learning from demand profiles and repositioning vehicles—preferably at night or at low traffic hours—will also play a role in overall service efficiency.
The same is true for personal pay-as-you-go vehicles like scooters and bicycles that can’t relocate themselves and may not have dedicated homes. Service crews will need to know where to reposition the vehicles each day for optimal utilization, and where and when to exchange flagging batteries with new ones.
It’s Not (Just) About the Vehicles
While many are fascinated (from fear or enthusiasm) about the advent of autonomous mobility, it has become clear that vehicles alone will not address the challenges faced by growing cities that already face near crippling traffic. By themselves, vehicles do not deliver a coordinated service that can move more people with fewer vehicles nor can they get the right vehicle to the right place at the right time. Services will need to be managed and optimized at the fleet level in order to be successful.