Johnson also said Waymo’s vehicles were increasingly facing pedestrians.
In one October Video, a Waymo car was driving through a crowded Costco parking lot. He waited patiently for them to be out of the way, then walked confidently.
“This number of pedestrians would have caused a use of the brakes causing a whiplash in March 2020,” Johnson wrote in an onscreen note. “And he would have given up completely in July 2019. Not anymore!”
The vehicles are still a little too careful with pedestrians. In recent videoJohnson called a Waymo vehicle to a crowded retail parking lot, pressed “start ride”, then had to wait almost 3 minutes before the vehicle moved a significant distance. There were apparently so many pedestrians and other vehicles around that the Waymo car did not feel safe moving forward.
A human driver would certainly have moved sooner. But it’s hard to blame Waymo for it – better to be a little slow than risk running someone over.
Of course four hours of perfect driving – or 40 or 400 hours for that matter – wouldn’t be enough to prove that Waymo cars are safe. Correctly assessing the safety of Waymo vehicles requires a lot of data. And Waymo has over 20 million miles of actual driving data. Almost all of that mileage is done on public roads with a safety driver behind the wheel. A small fraction – 65,000 miles until September 2020 – was entirely driverless.
Until recently, Waymo kept this data private, which made it difficult for the public to assess the technology. In October, Waymo took a big step towards greater transparency by publish data on the actual performance of its vehicles. It covered 6.1 million kilometers the company drove in the Phoenix metro area in 2019 with a safe driver behind the wheel, plus 65,000 kilometers of driverless operation from early 2019 through September 2020.
In 6 million kilometers of driving, Waymo vehicles have been involved in 18 accidents. Of course, for most of those miles, the vehicles had safety drivers who were supposed to intervene in the event of an impending collision. To estimate the performance of non-driverless safety cars, Waymo performed simulations of all situations where a safety driver took control. These simulations predicted that 29 more crashes would have occurred if safety drivers had not intervened.
While 47 crashes might seem like a lot, it’s important to remember the denominator. Waymo’s vehicles have been in an accident – or likely would have been in an accident without human intervention – once every 130,000 miles or so. This equates to over 10 years of driving for a typical human being who drives 1,000 miles per month.
It is surprisingly difficult to determine what the comparable rate would be for a typical human driver. Some of the 47 collisions reported by Waymo were extremely minor. For example, a pedestrian walked over the side of a stationary Waymo vehicle at 2.7 miles per hour. Two simulated accidents involved a bicycle and a skateboarder riding the sides of stationary Waymo vehicles at speeds of 2.2 and 5.9 miles per hour, respectively.
Such minor low-speed collisions would never be reported to police or other authorities, so we don’t know how many “crashes” like this one experiences a typical human driver.
More importantly, most of those 47 incidents appeared to be the fault of another driver. For example, one-third of real and simulated accidents were rear-end incidents. All but one – 14 actual collisions and one mock accident – involved another vehicle with a Waymo car. The last rear end was a mock crash where the Waymo car would have terminated the rear of another vehicle at a speed of 1 mph.