Reno, Nevada, may not seem like the place to develop the country’s first self-driving public bus, but Richard Kelley thinks it presents all the right challenges. The buildings are taller than those in the office parks of the Silicon Valley, providing a good visual test for the complex algorithms. The weather is more taxing, arid with occasional snowfall. Then there’s the foot traffic. “You have people who are just walking out of the casino,” says Kelley, the chief engineer at the University of Nevada at Reno’s Advanced Autonomous Systems Innovation Center. “They’re less predictable when they’re walking about.” If you’re teaching a 14-ton machine to navigate urban chaos by itself, Reno’s not a bad schoolyard.
On Tuesday, the University of Nevada announced the start of a three-phase project to get a real, live autonomous bus on the road by as early as 2019. The first stage starts June 1 when a sensor-laden, passenger-carrying electric bus built by California company Proterra, starts trawling a 3-mile route along busy Virginia Street. To start, a human driver will do all the work as the bus collects the data needed to navigate this first stretch. In stage two, researchers will use that info to build self-driving systems. By the third phase, they hope to commercialize and license the tech, and conquer even the craziest city streets.
You’ve probably heard the hype about self-driving cars. Though smart, rich companies like Google’s spinoff Waymo have spent years racking up millions of miles with the tech, robo-cars remain years from market-ready. Self-driving trucks promise to remake trucking.
But buses? Tesla CEO Elon Musk poked his head out of his tunnel to predict that buses will shrink and ditch fixed stops, opting for a Uber-like approach to pick ups and drop offs instead. As more algorithm-happy startups get into the transportation game, others are predicting public transit will go the way of the Model T.
Not so, says this Nevada collaboration that includes the local university, the regional transit agency, the state Department of Motor Vehicles, and Proterra. It has plenty of work ahead. Buses face all the technical obstacles cars do—intersections, cyclists, pedestrians—plus their own kinks. Buses have wider blind spots and must be trained using images taken from higher vantage points. Their cameras will have to grapple with the vibrations inherent in larger vehicles. Oh, and if something goes haywire, more lives are at stake.
But experiments like this show that robo-buses are coming, and that it’s time to grapple with what that means for public transit.
Bring on the Robot Bus
Public transit has many reasons to be interested in autonomous vehicles, and fortunately, only a few of them revolve around existential threats.
Number one: safety. Companies like Mercedes-Benz and Proterra already sell buses chock-a-block with the automated features you can find in most luxury cars: adaptive cruise control, lane assist, and warnings that beep when when people or vehicles enter drivers’ blind spots. This stuff works. A year-long study by the Washington State transit agency Pierce Transit followed seven buses sensored-out by the camera maker Mobileye. None of them hit pedestrian or cyclists, and drivers experienced 43 percent fewer pedestrian “near misses” than their non-assisted colleagues. The results so impressed the Federal Transit Administration that it gave Pierce Transit another $1.7 million to extend the automated features to the whole fleet. Automation just might keep everyone safer.
Agencies have another reason to get into autonomy: cold, hard cash. Buses are pretty darn safe—they kill 0.11 people per one billion passenger miles, compared to 7.28 in cars or light trucks. But they still injure people, and that costs serious money. Transportation planner and consultant Jerome Lutin found US transit agencies spent $4.1 billion on casualty and liability claims between 2002 and 2011. A safer public bus would mean cash for public coffers, too.
The Future of the Robot Bus
Now, you may wonder why anyone is looking at big ol’ buses. As tech companies perfect on-demand features, à la ridehailing apps like Lyft, Uber, and Via, public transit agencies are thinking deeply about providing door-to-door service, the bus minus the bus stop. To do that, agencies might need to deploy more vehicles, and zippier ones. Even the areas that stick with fixed routes might need to get smart about size. “Downsizing the vehicle based on the time of day and density of transportation makes sense,” says Susan Shaheen, a UC Berkeley civil engineer who studies mobility innovation. “Maybe at noon you don’t need a large-sized vehicle, so smaller vehicles could be available to provide flexibility midday.”
But even if some buses tend toward bijou, basic geometry tells you that dense cities will still need the big boys. “There are going to be certain corridors, even if the vehicles are automated, where you’re gong to have capacity constraints,” says Lutin. Market Street in San Francisco; 14th Street in New York; around the Loop in Chicago; and on Virginia Street in Reno: These roads get enough heavy traffic to make the human-hauling bus more efficient.
In fact, buses just might be the autonomy gateway drug, the way a fearful, touchy public gets cool with self-driving machines. “Many riders are already comfortable with subways, where you don’t interact with the driver,” says Matt Horton, the chief commercial officer for the electric bus company Proterra. What difference is a driver-free bus?
Well, years of study’s worth of difference, which is why the University of Nevada, Reno is embarking on this research now. Do not expect drivers to disappear from your bus commute any time soon, the region’s officials warn. “Maybe someday a bus driver will be more of a customer greeter, someone who is working closely with customers on their experiences,” says Joe Harrington, the public information officer at the Reno region’s transportation agency. In the meantime, the city’s got a lot of science to do.