Doing the Math: 2016 California Autonomous Vehicle Disengagement Reports

Self driving cars.  The imminent future of on-demand and super cheap transportation - or still a long development slog ahead, depending on who you listen to.  Earlier this year, California released the 2016 Autonomous Vehicle Disengagement Reports - which are reports of times that the human driver had to take over from the computers (either because the computers handed off control, or because the human wasn’t comfortable with what the computer was doing).

This is a companion discussion topic for the original entry at

(Comments from Blogger)

2017-04-17 by Unknown

Thanks for the insights, this is indeed some really interesting data to look at!

Short question on the BMW reporting: You listed “Disengagements per 1000 miles: 1.57” but isn’t the value actually ~1.74? I see it graphed properly, so maybe just a typo?


2017-04-17 by Unknown

Same with the GM data - isn’t the Disengagements per 1000 miles = 18.5? Just trying to understand how the numbers are calculated :slight_smile:

2017-04-17 by Russell Graves

For BMW, they had one disconnect in 638 total miles driven (over 2 months). So (1000 / 638) = 1.567…

The graph is showing the disconnect rate for April - note that they drove a few miles in March with no disconnects.

For GM Cruise, yes. That was a typo I’ve fixed.

2017-04-17 by Unknown

Oh yup, was looking straight in the report and missed that!

2017-05-14 by Niblick


Great detailed coverage and an honest look at the various AV testing programs.

While Waymo has the most number of AV mileage it is still a tiny speck of the 3.2 trillion miles we drive in the US on a yearly basis. Needless to say the rest of the players (GM, NIssan, Tesla, etc.) are way behind in this race. At this pace I honestly don’t feel we will see an AV commercially available in the next 5 years. What is your take?


2017-05-15 by Russell Graves

I think Waymo is the closest, obviously - they’re supposedly opening some self driving cars for family use in Arizona at some point. I think they can accomplish the bulk of tasks, in good weather (so, see Phoenix), soon.

For other companies, I’m not sure, but I wouldn’t bet on Tesla until 2019 at the earliest, and that’s a stretch.

And in terms of stuff I care about, like snow and trails? Hah. Nobody in Silicon Valley cares about dirt tracks.