Waymo's self-driving minivans have always had safety drivers as backups, whether or not the vehicle was a part of its public pilot. But that's about to change.
Waymo announced Tuesday that it will go completely driverless, with the eventual goal of launching an autonomous car-sharing service. Its fleet of Chrysler Pacifica minivans will be capable of SAE Level 4 autonomy, meaning they won't need a driver within a predetermined area. Waymo promises that, over time, this geofence will grow to encompass the entire Phoenix region, which is over 600 square miles and also the location of .
Ditching the driver
The Google spinoff already allows groups and individuals the chance to ride in Waymo's fleet of self-driving Chrysler Pacificas in the Phoenix area. But there's always been a safety driver riding up front, just in case something went awry. Now, Waymo's at the point where it believes it can ditch the driver entirely.
It reached this milestone with the help of simulation. Waymo's simulator allows the brains of its self-driving cars to practice driving on a scale that's a bit difficult to reproduce in meatspace -- it has roughly 25,000 iterations of its vehicles "driving" all day. Over the last year, Waymo has simulated over 4 billion kilometers of driving. At present, its vehicles cover 16,000 actual kilometers a day, and 1,000 times that much in simulation.
Waymo's big move will require an equal leap of faith from its passengers. To that end, Waymo put a lot of work into the UI embedded in every one of its vehicles. It displays pertinent information to its passengers, like the sorts of obstacles it's detecting on its drive. It color-codes pedestrians and bicyclists to prove to its human riders that it knows the difference between the two. It's the same UI that we experienced when we took a ride in Waymo's car.
In terms of redundancy, there's a whole bunch of it. The traditional systems like steering and braking have backups, and the same goes for the computers that power the vehicles' self-driving hardware. If there's a problem, the car can pull itself over, if need be.
Autonomy as a service
According to Waymo CEO John Krafcik, the eventual goal of this effort is to create a shared mobility service. He pointed out that most cars spend most of the time sitting idle, and a majority of trips in the US are only a bit over a mile. He believes Waymo's autonomy-as-a-service could take care of all that without the pain of actually owning a car.
"With each car being driven so little, and for what are mostly short trips, Waymo's technology allows vehicles to be used in a different way," Krafcik said in a speech in Lisbon. "A small fleet of fully self-driving cars could serve an entire community. There are other benefits, too. Parking lots could be transformed into parks; fewer traffic crashes could ease congestion."
Minivans are just the start. Krafcik envisions a whole ecosystem of these cars, where passengers can call on a vehicle that suits their individual needs, whether it's a big SUV for long weekend trips or something more compact for the daily commute.
A fully fleshed out service is still a ways off, and any sort of hiccups in this early pilot program could very well set the whole thing back by years -- the court of public opinion is a harsh one -- but Waymo's decision to remove a human failsafe marks a big step in introducing proper autonomy to the public.