Autonomous Chevy Bolt EVs drove 9776 miles in 2016

[marta]

The fleet of 20 autonomous GM Bolt Electric Vehicles drove almost 10,000 miles in 2016. The vehicles kicked out of automatic mode 181 times (they call it an 18.5 % drop out rate). I imagine that is when the vehicle encounters something it doesn't know how to handle?

Also of slight interest is the strange names that GM and Cruise Automation (the autonomous driving subsidiary of GM) gave the vehicles.

 

[JupiterMoon]

So Bolt drivers are going to have the pleasure of seeing their cars used as taxis and rental cars? Nice.

 

[Jychevyvolt]

The most impressive part was where the driving took place. Urban environment, the hardest part of autonomous driving.

 

[jjr]

Cruise Automation (the software company partnering with GM to work on autonomous driving) released video #2 of their Chevy Bolt EV (the one named Platypus I guess?)

https://youtu.be/Vfgjemwc9NU

 

[jjr]

Sorry just noticed at the start of the video it specifies 'Albatross' as the Bolt electric vehicle in the video.

 

[SeanNelson]

Cruise Automation (the software company partnering with GM to work on autonomous driving) released video #2 of their Chevy Bolt EV (the one named Platypus I guess?)

The video looks very impressive. But of course they're not going to release videos of rides that had problems. I wonder what the success-to-failure ratio is? And I wonder how the cars perform in inclement weather or at night?

 

[EldRick]

Well, that's about one percent of the miles driven by Google and Tesla autocars...

 

[Jychevyvolt]

Well, that's about one percent of the miles driven by Google and Tesla autocars...

Autopilot != Autonomous

 

[Nagorak]

What I've noticed in watching these autonomous driving videos including one on Teslas crash avoidance is that the computer is able to detect and avoid crashes and obstacles but a human who is aware of his surroundings can anticipate a problem much sooner. In the Tesla crash avoidance videos, several times I was able to see the problem way down the road before the car reacted. It sometimes came down to recognizing that a car might be thinking of changing lanes based on how the driver was acting--the sort of intuition which it is impossible to accurately proven. Here with the stopped delivery truck, a human recognizes the situation right away, but the car has to stop and think about it (in fairness I'm actually impressed it figured the situation out at all).

In the end I expect autonomous cars will be better than the average driver, because there are a lot of bad drivers out there, but it won't be nearly as good as an alert human driver. Of course, we all get tired and distracted, so maybe being 95% alert 100% of the time is ultimately better than being 100% alert 95% of the time.