The kids took a Waymo recently in San Francisco. Jillian and I drove to Seattle in January using Tesla’s Full Self Drive (Supervised), hereafter FSD, for about 90% of the trip in both highway and city settings.
Waymo uses an array of sensors to operate on a specific map. It’s been working in some major cities (Austin, Phoenix, San Francisco, etc.) for several years. Generally, it seems nowadays to work well. The car itself is a Jaguar iPace. The cost plus sensor array is around $200,000 per vehicle, making it roughly 4x as expensive as Tesla solution (which is just a standard Model Y with a new taxi app). My understanding is that right now Waymo loses $2-$3 billion a year.
Tesla is chasing a similar objective of Uber-like ride hailing in the short-term, and full autonomous vehicle ownership in the longer-term. The Tesla Model Y Jillian and I used was not the latest model available, though it had Hardware 4 (the latest computer). It was exceptionally good, handling four way stops with other cars, right turns on red, and other challenging driving scenarios without issue. At one point in the evening, it even swerved to avoid a cat that I guarantee I would have flattened had I been at the wheel. It was a remarkably good performance, with caveats.
Those caveats: It had difficulty entering and navigating in parking lots. I don’t know if this was it not knowing exactly where to go or some other factor, but these were the only points where I took over control so that we didn’t block traffic behind us. The car also had a tendency on freeways to want to override my preferences and speed up, well beyond the speed limit, to keep up with traffic (which was also speeding). It was aggravating to have to continually modify settings to get the car to slow down, but it didn’t require that I take control.
This experience of FSD was eye-opening. Despite being great at freeway driving—I’ve used that for over six years now—I’d long expressed skepticism about Tesla’s ability to get a car to handle all the variables of city driving. “It’s at least five years away” I’d always say. Well, I guess it’s been five years, because it was very good in January. That people are now having excellent Tesla RoboTaxi experiences in Austin, Texas based on stock Model Ys does not surprise me. Tesla may not have quite “solved for driving” yet, but they’re close when it comes to doing it within specific maps, and no one should be stunned if they actually have a quite viable automated Uber service running successfully across most US cities within the next couple years. Since Tesla’s vehicle cost is about a quarter of Waymo’s and ongoing costs do not include a human driver like Uber has, one can see the business case.
What this portends for the future of automobiles is a fascinating question. A review: https://www.youtube.com/watch?v=We2ZD0-IXPM