The California Department of Motor Vehicles have just released some new statistics on test autonomous vehicles (AV’s)  operating in that state. California requires owners of these vehicles to annually report the number of hours driven  and the number of hours that the vehicle is~well, steering itself. As reported in the Atlantic Monthly Waymo the self-driving car from Google owned Alphabet is still leading the pack. Waymo  AV’s “drove 352,545 autonomous miles with 63 total disengagements, for a yearly average of 5,595 miles per disengagement. In the company’s best month, November 2017, they did 30,516 miles with a single disengagement. ”
While these numbers are remarkable, it was noticed that the “disengagement rate” did not decline from the 2016 numbers. Waymo put two million total miles on their cars in twenty different cities in the last year. In 2018 Waymo is commencing a “self-driving taxi service” in Phoenix, a trial that will expand to other cities. Waymo uses customized minivans produced by Chrysler and it is expected that thousands of these AV’s will be produced as Waymo continues expanding to a truly autonomous service.
Who else is in the field? The number two AV tester in California,  GM’s Cruise, drove 125,000 miles in San Francisco in self-driving mode, with 105 disengagements. Tesla reported zero miles. Instead of using costly sensor arrays Tesla is arguing that their “billions of miles of real-world driving data” will develop their autonomous technology. The “shadow mode” in the Tesla gleans data without actually taking control of the vehicle, developing the system by data accumulation as opposed to physical vehicle testing.
As the Atlantic Monthly writer Alexis Madrigal concludes: “We’re not seeing a spike in the number of companies that are able to execute well on California roads, nor are we seeing an explosion in the number of miles driven or massive reductions in disengagements per mile. It could be that this is happening in other states, outside the regulatory reporting spotlight. But for now, this looks like a two-car race with a Tesla revving its engine in the infield.”


  1. “Disengagements” is a fancy way of saying “The car got in to a situation it couldn’t get out of without human intervention”. We currently talk about this “human” intervention as needing to come from an occupant in the car; likewise it seems said intervention is the major reason we can’t go to full driverless, have a nap, ride while intoxicated, etc.
    My guess is we’ll see “autonomous car control centers” pop up, the same as we have for military drones. A drone pilot intervenes under specific circumstances, but the drone takes care of itself for the most part. Imagine a call center full of folks who’s job it is to simply navigate your “stuck” self driving car out of whatever situation it has gotten itself in to…
    That would actually make it quite possible to move forward almost today based on the technology stats in the report…

    1. That’s hilarious, people allowing some kid in a distant call centre take control of their car. While they’re in it. Like you, I envision evolution of full logic technology; after we find too much human error in the call centres, we can replace the centre staff with ACCESS (Algorithmic Control Ctr Emergency Support Systems), artificial intelligence that engages the disengaged vehicle.

  2. That’s what, roughly 8000k per disengagement?
    Sounds OK at first glance, but if you think about it, there’s maybe 750,000 people who commute by car in Metro Vancouver, and they travel, say 40k round trip in a day on average, that is 30 million km per day during rush hour.
    At one disengagement per 8000k, that is 3,750 disengagements during rush hour every day. Imagine what happens in bad/unusual weather! Either the cars need to get better by at least 2 orders of magnitude, or I’m buying shares in whoever owns AM730 or whoever is building Bryn’s call centers.

  3. I’m not worried about driverless cars simply because they’re not going to happen. To my eyes it looks like a scam (by an industry that isn’t doing well) to try to hang on to shareholders by promising a future where cars will still be as relevant as the past.

Leave a Reply

Your email address will not be published. Required fields are marked *