Waymo’s self-driving robotaxis can successfully nail a tricky left turn, weave through lanes to drop you off at the airport, and safely pass a U-Haul that’s idling in the middle of the street. But during a blackout, they apparently turn into four-wheel bricks.
On Saturday, when a major power outage in San Francisco knocked out traffic signals, many Waymo vehicles didn’t pull over to the side of the road or seek out a parking space. Nor did they treat intersections as four-way stops, as a human would have. Instead, they just … sat there with their hazard lights on, like a student driver freezing up before their big parallel-parking test. Several Waymo vehicles got stuck in the middle of a busy intersection, causing a traffic jam. Another robotaxi blocked a city bus. The company suspended its service for several hours.
Waymo vehicles should be able to call up a human agent for a remote response if they get confused or are unable to operate. But connectivity issues during the blackout may have prevented this safeguard from working. The company is diagnosing the problem, a Waymo spokesperson told me in a statement: “While the failure of the utility infrastructure was significant, we are committed to ensuring our technology adjusts to traffic flow during such events.” The spokesperson added that the driverless cars should have been able to navigate intersections without working traffic lights, but “the sheer scale of the outage led to instances where vehicles remained stationary longer than usual to confirm the state of the affected intersections.”
This isn’t the first time things have gone awry with Waymo in San Francisco. One of the company’s robotaxis got stuck in a construction site. Another fatally struck a beloved bodega cat, and there has been at least one incident in which multiple Waymo vehicles got into a standoff where no car seemed sure how to proceed. But without a doubt, last weekend’s incident is Waymo’s biggest crisis yet.
[Read: San Francisco’s nocturnal taxi ballet]
The company hopes to bring its fleet of fully autonomous Jaguar SUVs into many more city streets in the coming years. The cars—powered by lidar, cameras, and other sensors—are remarkably safe, logging fewer crashes over more than 125 million miles than the average human driver on the same roads. And although Waymo has been involved in two known fatal crashes, both were found to have been caused by human drivers in other cars.
Waymo has parlayed this safety record into an enormous expansion. You can already ride in a Waymo in the Bay Area, Phoenix, Los Angeles, Atlanta, and Austin, and the company’s cars are set to hit the streets in many more places: London, New York City, and Washington, D.C., are next, provided that local politicians and regulators are fully on board. But the blackout is a stark reminder that even as driverless cars go global, we’re all taking part in their public beta. As Waymo enters new cities, more opportunities arise for things to go wrong. Nasty winters could affect how the cars’ sensors and cameras see and understand the road. Unique on-ramps or other unusual road designs could prove challenging to navigate.
A Waymo taxi may do great in normal driving, but it’s these kinds of edge cases where things get tricky—and where the consequences could be monumental. What if hundreds of Waymo taxis shut down at once during a San Francisco earthquake? How are human drivers, or first responders, supposed to communicate with them on the road? What if they blocked emergency vehicles during a Los Angeles wildfire? And how would they handle the chaos in New York during extremely unpredictable and unlikely events, such as the Knicks winning the NBA Finals?
Other driverless-car services face similar challenges. Tesla is building up its own robotaxi service, starting in Austin; Zoox, a robotaxi company owned by Amazon, has started picking up passengers in parts of San Francisco and aims to expand to other cities next year. Driverless cars have enormous potential. Nearly 40,000 people died in traffic crashes in America last year; the ultimate goal of autonomous cars should be to bring that number as close to zero as possible. But the technology’s promise means nothing if Americans can’t predict how these cars are going to act when the power goes out or when there’s some other unforeseen disaster. Maybe it’s impossible to anticipate all of the ways that things could go wrong. But if Waymo and its competitors want to prove that driverless cars really are better than all human driver, someone is going to have to try.
The post That Was a Lot of Frozen Waymos appeared first on The Atlantic.
