Recent Waymo announcements are slightly underwhelming


Waymo recently announced two new partnerships for their fleet of robotaxis.

The first was with Walmart. Walmart has installed special parking spots in their lots, and will offer people free rides to Walmart to pick up online orders. Only some of the 400 Waymo "Early Riders" will participate, at first.

The other project involves last mile rides to some stations of the Valley Metro transit agency. This begins only with the transit employees, but will expand to passengers of "Ride Choice" a special heavily discounted taxi service for seniors and the disabled.

The far more interesting recent piece is this Bloomberg interview with a teen-aged Early Rider about her experiences. Indeed, learning just how people interact with the cars is one of the key goals of a project like this.

While I'm obviously a Waymo fan (and worked on the project in its early years) I must say these announcements are a bit underwhelming, compared to the big milestone which Waymo has teased for 2018 -- rides for the general public, with no safety driver. It is disappointing, though understandable, that Waymo is doing these baby steps. The reality has always been that making a public service is a very hard problem. Waymo has a significant lead on the problem but nobody is fully there yet. I do believe we'll see it happen in Arizona and a few other places this decade, but in some places it is still several years away.

The Walmart project seems particularly unexciting. In fact, since the ride is just to pick up groceries ordered online, it's hard to imagine why this isn't more a task for delivery robots. (I work in the delivery robot space with Starship Technologies.) The need for special parking spaces, instead of just taking people to the door, suggests that such navigation at the busy door of a Walmart may be on the more challenging list.

The transit project elicits several thoughts. I have always felt that paratransit is a fantastic use of robocars. Paratransit (transit service for the disabled, technically "on demand") is terrible in the USA. You have to book it a day in advance and wait up to an hour, and it ends up costing cities an average of $30 per ride, nationwide. For the disabled, driving a car can be difficult or impossible, and even riding accessible transit can still be a pain. If anybody needs robocars with cheap taxi rides, it's this community.

On the other hand, I feel that "last mile" applications are wimpy. While they certainly make transit more tolerable, direct service along the shortest door to door route should be what this is all about. As I outline in The Future of Transit this can still be sharing a vehicle, and still involve transfers (but with no waiting or deviation from your route) but the old ideas of lines and schedules and stops should stay in the 20th century.

This will however, be the closest thing to a service for members of the general public.

Another underwhelming revelation comes in the interview with the Early Rider. While Waymo is operating vehicles without safety drivers, she reports having only ridden once in such a vehicle even though she uses it very regularly. This tells us that Waymo's deployment of no-safety-driver operation is still quite limited, and probably only in a limited set of conditions and routes. This again makes sense, but should temper our enthusiasm for Waymo's progress slightly, though its lead is still almost overwhelming. Nobody else is close to operations without a safety driver on open roads.

I think that Uber's fatality says almost nothing about the quality of Waymo's product. But a hard PR reality is that another incident in the same town will not be treated that way by the public, and Uber has made it harder for everybody. This could explain a scale-back on Waymo's part in operations without safety drivers. Uber effectively used a lot of the tolerance the public might have for the incidents which are going to happen -- even minor ones.


I was baffled by the notion of driving people to Walmart to pick up groceries they'd already ordered. One might as well use a driverless car to go to the post office to pick up your mail. Is there any good reason for doing this, other than PR and perhaps the hope that people might buy some extra stuff while collecting their order?

I think it's just a prototype test project. Possibly a means to experiments in delivery.

I agree. I have a couple thoughts here:

Waymo has spent a billion dollars+ on the project. If something happens now and there's not a human to take over, they risk not being able to launch at all, ala Uber. Something will happen, it's just a matter of when. The cost to have safety drivers in the cars for the next few months until actual launch is minimal and could actually save the project while giving them someone that can give the car feedback when it doesn't make a lane change aggressive enough or does something strange.

My second thought is that Waymo is likely confined by the amount of cars they are getting. Chrysler delayed the Pacifica Hybrid over a year because of issues and recalls and almost scrapped the whole thing. I believe that Waymo is year+ behind just because of this issue, which is why the announcements are so lame and scaled back. Reports are they have 160 cars as of a couple weeks ago. They obviously can't scale and announce a big launch that is open to the public without Chrysler really picking up the pace on the vans. That's said to come in November.

I do think that while Waymo is in an entirely different situation from Uber, the Uber fatality has to put fear into everybody. Not that everybody wasn't already afraid of the risk of this, but there is no question that if it happens twice in the same town that public reaction will not treat them as independent events -- whether they are or not. So I can understand scaleback.

If Waymo were looking for a way to make use of their technology right now and in a way that is transformative if not revolutionary, they could apply it to vastly improving the capacity of busways.
As you have pointed out before getting autonomous vehicles to deal with the controlled environment of a busway is a far easier problem to solve than fully autonomous vehicles.
You would keep the driver for ordinary roads, and use the self driving capability only for acheiving high density on the busway.
While this is not the ultimate goal at least it might prevent the spending of vast quantities of money on white elephant rail projects.
As to whether a detour like this would slow the development of fully autonomous vehicles, either through consuming Waymo engineering bandwidth, or by giving credence to the idea that autonomous vehicles require new infrastructure, I am not sure.

BTW, while Canada may or may not be the best country in the world, I am pleased to see them pissing off the Saudis. I won't be at all sad when the wane if the oil dollar lessens their influence.

I think that Waymo probably has full focus on real road driving for cars and trucks, and any distraction from that is probably an error unless it can do it in a completely independent team.

Platooning is something other companies, like Pelaton, Volvo and others, have already done. You don't need platooning, though, to make a busway better. Every dedicated busway I have seen has huge headways measured in minutes, not fractions of a second. You can get down to 2 second headways without any platooning.

I have an article in the queue about how to turn a subway into a super-efficient van/busway.

I would view these as political coalition-building efforts. The potential political backlash involved in displacing the economics of transport is going to require more than just better technology.

Add new comment