But what about self-driving taxis?

I wrote an earlier note about how self-driving cars are not happening. So why do we still read about self-driving taxi trials in San Francisco and Phoenix? It is very interesting to me that the two cities that are running trials of this technology are so different, and both pretty familiar to me.

Phoenix is a downtown that is pretty barren of both pedestrian and vehicular traffic, and the surrounding areas are even more sparse, with the possible exception of the student population in Tempe. A poor, confused and lonely autonomous cab can just pull over to the side of the road without being noticed until its handlers arrive to service it.

San Francisco is the other end of the spectrum. There is a lot of pedestrian traffic, a lot of vehicle traffic, the road configurations and restrictions are crazy, and forget about finding street parking. The pedestrians are in charge of a great deal of downtown, bicyclists are aggressive and sometimes travel en masse, and there are many different types of buses, trolleys and cable cars. I hate to drive in San Francisco.

It might seem curious that while we see companies backing off claims of self-driving cars for the foreseeable future, self-driving taxi services are seeking to expand. But on closer examination, it makes sense that self-driving technology would best be proven by taxis first. Self-driving taxis are a subset of the general self-driving car problem, for at least three reasons:

1. They can limit their range to a very well-known set of roads.
2. They can operate a fleet that is maintained to rigorous standards.
3. They can monitor their behavior closely, albeit remotely.

The following New York Times article explains where this effort has gotten to, after more than ten years of research and development.

Self-Driving Car Services Want to Expand in San Francisco Despite Recent Hiccups

Waymo and Cruise hope to widen their projects to more of the city. But local officials are increasingly concerned about breakdowns causing congestion.

Note that the software guys are still talking of “edge cases”. A quote from the article:

“Sometimes these cars just need a human to help them out of a tough spot” – Phil Koopman, an engineering professor at Carnegie Mellon University who specializes in autonomous vehicles.

Listen pal, I’d like to help you out of this tough spot, but that’s not what autonomy is all about.