Self-driving cars are such a common sight in San Francisco that they’re hardly worth a second look. But when an autonomous vehicle — sans backup driver — gets pulled over by the police, that’s worth a gander.
And that’s what happened the night of April 1, when a San Francisco police officer pulled over a fully autonomous Cruise vehicle in the Richmond District.
Despite written plans to deal with these kinds of scenarios, the video suggests that neither the police, nor the autonomous vehicle, were particularly well prepared for this encounter.
The video begins with a police officer approaching a stopped Cruise vehicle, seemingly unsure of what to do. As the officer was walking back to his cruiser, the AV surged forward, crossed an intersection and pulled over at the next block. The police cruiser followed close behind. The Cruise, a modified Chevy Bolt, then remained stationary, as police officers spent over a minute trying to gain entrance to the vehicle. By the time the video cuts out, the vehicle had rolled down its windows and a police officer was talking on the phone.
According to Cruise, which is owned by General Motors, the vehicle was pulled over due to a failure to turn on its headlights — an issue the company says it has corrected. The vehicle drove forward after initially being pulled over in order to stop in a safe location, the company said.
A Cruise spokesperson said that police officers on the scene contacted the company, and that the vehicle was released without a citation.
State regulations require autonomous vehicle companies to provide “law enforcement interaction plans” that explain to police officers how to engage with these vehicles. The April 1 incident — which was not a stunt or joke — represents an early test of how these plans work in practice.
Cruise has conducted trainings and walk-throughs with the San Francisco Police Department, the fire department and California Highway Patrol, a spokesperson said.
SFPD did not respond to questions about how AV law enforcement interaction plans are disseminated among officers, or whether every officer is briefed on these plans.
The Examiner reviewed Cruise’s law enforcement interaction plan as part of an investigation into AV regulation published last week.
The document, which is available for public viewing on Cruise’s website, says that its vehicles are programmed to identify first responders, and to safely pull over when instructed to do so. First responders must call a 24-hour hotline to gain access to the vehicle. Dashboard icons indicate whether the vehicle is still in fully autonomous mode. If need be, Cruise agents can direct first responders to disengage autonomous mode and take over from the computer to move the vehicle.
Passengers riding in the back seat of a Cruise AV have access to a two-way communications link with Cruise staff and an “end ride” button. Cruise staff will be notified immediately in the event of a collision and will get in touch with passengers.
Waymo, another AV company doing driverless testing in San Francisco, has similar policies as part of its law enforcement interaction plan.
The April Fools’ Day incident comes as Cruise and Waymo have ramped up their testing and ride-hailing pilots on the streets of San Francisco.
Between December and February, fully autonomous Cruise vehicles drove more than 3,300 miles as part of its driverless ride-hailing pilot in more than 750 rides, according to data from the California Public Utilities Commission.
Waymo, which began its driverless ride hail pilot for employees in March, has been providing rides to select members of the public in autonomous vehicles with a backup driver for several months. Between December and February, its drivered ride-hailing vehicles logged 911,000 miles in more than 8,300 rides, according to CPUC data.
That’s on top of the more than 3 million miles — more than six round trips to the moon — that Cruise and Waymo drove, mostly on San Francisco’s streets, in 2021, representing more than 75% of miles traveled by AVs in California.