We’re all going to be jetting around taking naps and playing parcheesi in self-driving taxis by next year, right?
Wrong. We aren't even close.
The results of Uber's massive self-driving car experiment are in, and things didn't go quite as smoothly as expected. They measured a ride’s success in a few ways: the average number of miles a car drove itself before a driver had to take over for any reason; the average number of miles between “critical” interventions when a driver had to avoid harm (like hitting a person); and the average number of miles between “bad experiences,” which is simple a measure of jerkiness and comfort level.
Let's just say thank goodness there was a driver.
Over 20,354 miles logged, there were 20,354 instances of a driver having to take control of the vehicle. That's one intervention per mile, up from one intervention every .8 miles on their previous test. To boot, the rider experience overall has been poor, with pestering braking and “bad events.”
It appears that self-driving tech is imperative to Uber’s future profitability: it's no secret that the company is currently hemhorraging money, and computers are cheaper than drivers in the long run. The company has also had a recent string of PR calamities that included groping, cocaine use and the CEO just being kind of a dick, so maybe people in general just aren't really their strong suit.
For now, it appears robots aren't, either.