Waymo has begun giving rides to the public on freeways in the San Francisco Bay Area, Los Angeles and Phoenix, and expanded its Bay Area service all the way down to San Jose, adding SJC San Jose Airport. This long-awaited expansion will greatly improve the service and shorten many rides.
While Waymo has been operating on freeways for employees for quite some time, the limitation to city streets has been a frustration to riders, since many trips are quite a bit longer without the use of the freeway. The limitation has seemed odd, since freeeway driving is much simpler than city street driving, and many robocar projects started freeway only because of that simplicity. Everything from adaptive cruise controls and Tesla Autopilot to Aurora trucking to Mercedes Drive Pilot (which allows drivers to take their eyes off the road in special limited situations) are freeway primarily (or freeway only) or began that way. Even Waymo began that way.
On the freeway there are (almost) no pedestrians or cyclists, no oncoming traffic, no traffic lights or intersections or crosswalks or so many of the things which complicate street driving. What’s scary, however, is speed. The destructive energy of a crash goes up with the square of speed. A crash at 70mph has 8 times the energy of one at 25mph. Braking distances are also much longer, reaction times must be quicker. If you do crash at low speed, you may only see property damage. At freeway speed, you will often see injuries, or even death.
In addition to speed, freeways can have lane-splitting motorcycles, hydroplaning and you need longer sensor range. You may encounter high speed crashes, fast cut-ins and unusual debris.
Vehicles will use the freeway only when it offers a meaningful time savings. In addition, riders must enroll in Waymo’s “trusted tester” program in the app in order to be taken on the freeway at first. Vehicles will limit themselves to the speed limit (which will make them the slowest vehicles on the SF Bay Area’s Interstate 280.) Operation will be 24/7 though very heavy rains may make the cars prefer city streets. If a user asks to “pull over” on freeway, the car will take the next exit, first.
Waymo waited to reach even higher levels of safety and now conclude they are ready. Freeway driving is also usually a must for airport trips, which are a very important part of the ride-hail business that Waymo currently finds itself in. SJC airport already has a designated robocar drop-off spot (which Tesla also plans to use with their safety-driver supervised pilot,) annoyingly a bit further than the Uber/Lyft spots. According to reports the airport will charge $4 to the company for each ride.
(This charge is counter-productive, since robotaxis will be better behaved than the private cars dropping off passengers at no charge. The private cars get to go right to the terminal curb. The robotaxis can, if things sync up, pick-up a passenger right after dropping one off, halving the traffic into the airport. They should be rewarded, not punished.)
No airport drop-offs at LAX for now. LAX planned to open their people-mover to take passengers to the ride-hail pick-up lot in January, but has pushed it later in the year.
The service expansion will add a great deal more territory–including, I am pleased to say, my own home. Most of Silicon Valley, the home of Waymo and Google and many other robocar projects will now get access. Unfortunately, that valley is very car oriented, with poor transit, plentiful free parking and an estimated 95% car ownership. Few people there need a taxi, except for things like trips to the airport.
In Phoenix, Waymo recently allowed teens 14-17 to ride on their own. Since an estimated 69% of schoolchildren in the valley (yikes!) are driven to school by their parents, it’s easy to see wealthier parents giving the job to a Waymo when that’s turned on.
Freeway Crashes
As noted, one key reason to delay freeway deployment is the extra dangers of freeway speeds, including the risk of severe injuries and death. Waymo co-CEO Tekedra Mawakana recently stated that no vehicle will be perfect, and so such crashes will eventually happen, but that she believes society will accept them, so long as the vehicles maintain an overall superior safety record. Because the public focuses on events rather than statistics, any such event will be a time of tumult. Uber ATG caused a fatality (though due to safety driver negligence) and was eventually sold off to Aurora. Cruise caused a serious pedestrian injury and was shut down. Waymo recently was involved in a fatality when a motorcyclist rear-ended a Waymo that was slowing for pedestrians, and bounced into the next lane where another car fatally struck the biker and fled the scene. In this last incident, there was no uproar due to lack of fault by the Waymo.
So far, Waymos have been having a lot fewer crashes and injuries that human drivers on the same roads, and Waymo obviously feels their testing indicates they will maintain that record on the freeway. But they have had, and will continue to have incidents, which now may be more severe. When the first incidents come, Waymo will need to help the public and regulators understand the benefits of the reduced risk from their vehicles in light of imperfection. The public is fickle though. A Waymo recently hit and killed a bodega cat in San Francisco, creating calls to take Waymos off the streets and even one non-binding resolution towards that end from a city supervisor. Waymos have, based on NHTSA reports, hit perhaps 5 cats and killed 2-3 in 100 million miles of operation, while statistics suggest humans put back behind the wheel by a robotaxi ban would have killed over 150 cats. No cat lover should want to pull Waymos from the street for that, and if they maintain even a fraction of that record with freeway crashes, it would not make sense to react to them with efforts to ban them.
Conquering the World
Waymo now has a much more complete service, with all types of roads over a large area. The also plan expansion to places with snow, and very busy streets like Manhattan. They recently reported doing one million rides in the prior month in California, and combined with other cities they must be approaching 350,000 rides/week or more. With freeways they will now accumulate miles at a faster rate. Baidu has operated on freeways in China, and others have operated with supervising safety drivers, but that means fairly little compared to this big step. Waymo is now ready to move towards full scaling, and replacing car ownership for some customers. (Though in the Bay Area, they will need to add some of Marin and the East Bay as well as downtown San Jose and SFO, which they are testing.) Only Baidu is close, and they can’t legally operate in the USA.