The Uber that blew through a red light in San Francisco raised unanswered questions about policing, manufacturers’ guidelines, and shared space.
The video, dated December 14, now has close to two million views on YouTube: The light turns red at a busy San Francisco intersection, a pedestrian steps into the crosswalk, and an SUV with a giant headlamp sails right through. It was an Uber equipped for self-driving, the sort of vehicle that’s promised to essentially eliminate the dangers of human driving. But it had just narrowly avoided collision with a person exercising his right of way.
Tech companies are afforded a lot of special privileges in their use of San Francisco’s public space, but they’re also subject to heavy scrutiny. It didn’t take long for officials, and the media, to discover that Uber, which had started testing robo-Volvos with humans behind the wheel in San Francisco days before, wasn’t actually permitted to operate autonomous vehicles in California. Uber claimed that the red-light fiasco was caused by human error, and that they hadn’t needed the permits since the cars had human drivers monitoring the wheel. California wouldn’t hear it: The DMV ordered the cars off the roads, or else—so Uber moved AV testing to Arizona, where the governor is thrilled to have them.
Clearly, Uber’s hail-able AVs won’t be withering away under the Southwestern sun. The San Francisco incident raised a lot of questions about how new mobility services and automation intersect, and what lies ahead for cities working out how to handle the promise of shared, self-driving vehicles. Here are three areas where we can expect more action soon.
License and registration, please
How do you ticket an AV that messes with the law? After the Uber incident in San Francisco, local police officers said they were unaware that the company had even started testing on city streets. But officers said that if they see AVs operating in violation of California vehicle code—which requires passengers capable of taking over the vehicle in case of malfunctions—they’d have enforced those laws. How would the SFPD respond to an autonomous Uber cruising through a stoplight? “I don’t even know,” Officer Giselle Talkoff told reporters. “I guess we could pull them over.” Sergeant Vince Lewis with the Phoenix Police Department, which is about to start seeing Uber’s AVs hit the road, sounded much warier.
Joseph Schafer, an expert in technology and policing at Southern Illinois University, says it’s almost by necessity that police departments aren’t really prepared for the autonomous present, let alone future. “There is work going on with some of the major police professional groups to get ready, but this is a good example, and not the last where we’ll see, where the ‘getting ready’ isn’t pushing down fast enough to personnel dealing with these vehicles, infractions, and questions,” he says.
Part of the problem is that critical questions of liability have yet to be worked out: For example, is Uber responsible for infractions on the part of drivers minding AVs, or is that the province of the drivers themselves? In a few years, when fully autonomous cars are driving on city streets, will it be car manufacturers or car owners responsible for run-ins? AVs promise to slash road fatalities, but they can’t eliminate them entirely. Google, Mercedes, and Volvo (which makes Uber’s AVs) have all stated they are liable for damages “from accidents their self-driving cars cause while in autonomous mode,” Slate reported earlier this year. Some believe manufacturers should be held responsible for traffic violations caused by their AV technology as well.
Meanwhile, what state laws exist mostly place liability on the person behind the steering wheel or, if they’re in the backseat, whoever flipped the AV switch. It’s going to take more testing, and likely, more collisions and more lawsuits, to work out these uncertainties and others—and for police departments to find their (perhaps increasingly diminished) role in enforcement.
Making up the rules
Some experts think that we shouldn’t spend too much time worrying about policing AVs, though. “Once companies know what they need to in order to get a license, so to speak, it’ll be easier, and there won’t be as much of a need for policing,” says Hod Lipson, an AV expert and roboticist at Columbia University.
Lipson is the co-author of a new book, Driverless, which makes the argument that federal regulators need to be much, much firmer in setting the bar on safety requirements for manufacturers and software companies. In September, the U.S. DOT roughly sketched out a 15-point AV safety assessment, its very first set of comprehensive guidelines for manufacturers. It puts forth generalized, non-technical standards on big questions of passenger safety, like what happens when autonomous technology fails, and how manufacturers are supposed to share collected vehicle data.
The DOT guidelines were applauded by many in the industry for being flexible enough to accommodate the fast-changing technology. But Lipson says that manufacturers are still hungry for cut-and-dry rules on proving a car’s safety. “It should be like when the FDA asks a manufacturer to prove that the drug is safe,” he says. “The DOT should come at it from that point of view: require companies to [provide] convincing data that the car is twice as safe as the human driver, and otherwise stay out of their way.”
The feds do plan to refine their guidelines as technology evolves. But from the DOT’s perspective, states also need to step up to make their rules on AVs more consistent with the federal guidance. So far, just eight states have passed legislation related to self-driving cars, while governors in two others have issued related executive orders. Some, like California, require human operators capable of taking over the AV (though California has also proposed legislation eliminating this requirement, which would align the state with federal guidance); others, like Michigan, don’t require humans ready to spring into action. Other states, like North Dakota, haven’t gone further than provisioning the “study” of the technology. The other 40 states have no self-driving laws at all, even while some are still testing the cars—leaving open a big window on what’s legal and what’s not.
There are pluses and minuses to letting the states have control over what’s allowed within their borders. It could encourage more and nimbler experimentation with law-making. Then again, manufacturers don’t want to have to design different cars for every state. Lipson says that when there are clearer rules that apply across the board, real-world testing will begin to take off at a much higher velocity. The industry will begin to generate the data it needs to refine the technology and improve safety. “It’s chicken and egg right now,” he says. “These issues”—he means like Uber’s San Francisco run-in—“will go away once there is clear criteria.” Expect to see more states attempting to legislate such criteria in 2017 and and beyond.
Curb’s the word
Between the parked cars and moving buses (not to mention bike lanes) that already fight for the territory between street and sidewalk, curb space is a hot commodity, and only stands to get hotter. Shared mobility services of all kinds—ride-hailing companies, micro-transit shuttles, carshare parking, bikeshare stands—usually require some slice of curb to operate smoothly. Vehicle charging stations will also start elbowing their way in as electric vehicle adoption spreads. The curbs of the future may also include special “docking” areas for passengers to load into shared autonomous vehicles.
“When you look at the highly automated future”—which seems to be the business model Uber is after, given its aggressive testing (Uber did not return request for comment as of press time)—“there will be a need for access to infrastructure,” says Susan Shaheen, a shared mobility expert with the University of California, Berkeley, who has done research with Uber, Lyft, Car2Go, and other mobility companies. “How do we as a society prioritize those curbs, those rights of way, those docking areas needed to fill, recharge, and maintain these vehicles?”
The answers will profoundly inform the future of urban transportation, and cities should work actively to find them. Perhaps a city will want to prioritize curb access for high-occupancy vehicles. Maybe companies will need to pay more during peak periods in areas where they compete with traditional transit services. Public officials will need to take the lead on getting shared autonomous vehicles into lower-income neighborhoods. Rethinking the curb is going to take a lot of buy-in from not only city leaders but also from retailers, who are often unhappy about losing storefront parking spots. These complicated conversations are slowly beginning to happen among transportation planners, says Shaheen, but she hopes to see a lot more.
Uber’s incident in San Francisco didn’t directly hit on the question of curb-space, but it did seem to signal to the city (as it has to so many others) a lack of interest in playing by the rules. Some cities—like some of those that have already set up partnerships that essentially replace transit services with subsidized Uber rides—may be thrilled to sell off a portion of curb space to the first company that wants to pay. But in the densest cities, mass transit isn’t likely to go away any time soon, and lots of companies are vying for the same court-side seats, as it were. In those settings, companies that flout the law might find it difficult to acquire the slice of concrete they desire. As with all things autonomous, watch this space.