Driverless legislation pushed

[Jflem1983]

Well there is always expert opinion like above.

Or there is this.


https://www.castanet.net/news/World/221 ... r-scrutiny

[Jlabute]

Since they are making the analysis public, I am looking forward to their assessment of what happened in the cars brain. Obviously the car saw and categorized the person. Maybe 'AI' has already begun to peg people off.
We don't expect that this will slow down the development of the technology, we are just saying these cars won't be flying off the shelf in a few years. There are a number of things that have to happen to bring LEVEL 5 vehicles to your local dealership... the electronics can't add $20,000 on top of the price, should work in all weather conditions with minimal maintenance, can't be so horribly ugly, the vehicle won't come with one or two engineers, and last but not least; can't kill people and you can prove it won't. Maybe this car had AlphaGo AI tech in it... so the more people it runs over, the better it gets.

Anyone having second thoughts about crossing the road?

[Jonrox]

The video of the crash has been released. I’m not sure a human driver would have been able to avoid her either. But the car should have been able to identify her better than a human driver could given it was completely dark and she didn’t show up in the vehicle’s headlights until the very last second (it looks like she was walking with her bike in a completely unlit part of the road).

[Jlabute]

I’m not sure what the relevance of her being lit up by vehicle headlights is, perhaps that may have helped the driver.. LiDAR should have spotted her 200 ft out or more in total darkness, shadows or not, 360 degrees around the vehicle. Radar and or cameras should have sensed her close up when all else fails. The video shows a complete failure of UBERs technology and safety protocols. Perhaps their software just sucks... I trust the sensors more than the software. UBER may release Lidar data soon.

https://www.theguardian.com/technology/2018/mar/22/video-released-of-uber-self-driving-crash-that-killed-woman-in-arizona

Who knows if this may have any bearing on Waymo since UBER and Waymo are in a lawsuit over UBER swiping Lidar IP from google.

[Jonrox]

My point about the headlights was only that there's a good chance a driver in that car would have struck her given how little time there was when she showed up in the headlights. I agree that the car's technology should have seen her though.

Acknowledging the car should have been able to see better than a human could is in fact an argument for improving the technology rather than shutting it down. We may not be there yet, but technology will soon result in a safer autonomous driver than a human could ever hope to be.

[Jlabute]

So now the truth is coming out. We won't be seeing fully autonomous (driverless) driving vehicles on our car lots come 2020 as many had thought, but now it is a minimum of 10 years out. Makes sense seeing what's happened in the last 5 years even with drivers in the vehicle. Personally, I think the estimate of 10 years is very naive and I would guess 20 (more than the author). Makes you wonder who is speaking on behalf of the engineers.

"Manufacturers are now admitting that full autonomy of vehicles is at least 10 years away. "

"All of the enthusiasm of a few years ago has somewhat petered out as manufacturers dig deep and invest more in R&D to move vehicle autonomy forward."

https://www.castanet.net/edition/news-story-277600-1102-.htm#277600

https://www.vox.com/future-perfect/2020/2/14/21063487/self-driving-cars-autonomous-vehicles-waymo-cruise-uber