5 ms·

2-3 years in OK conditions, 30 years to cater for all the odd edge cases humans take cars. Bad human drivers, Indian traffic, cyclists, animals, narrow, unlit o

2-3 years in OK conditions, 30 years to cater for all the odd edge cases humans take cars. Bad human drivers, Indian traffic, cyclists, animals, narrow, unlit or poor roads etc. The other question is will the tech be there before the licensing/legislation. Somewhere in the middle there's probably a point where self driving is demonstrably safer than human, even before they're "ready" for everywhere.

petra 10y ago
For a deep learning machine with LIDAR , there's only small difference between animals, indian traffic, bad drivers, and narrow/unlit/poor roads. It's just a question of more training(altough a recent NHTSA? regulation says the planning stage cannot use black box techniques like DNNs). And once self-driving is a business, i suspect we'll find ways to get said training data for cheap enough and fast enough. For example Tesla/Uber drivers as trainers .
SEJeff 10y ago
You're grotesquely underestimating what fleet learning will do for traffic issues. The way I see it (as a software engineer myself), is that each wreck that the AP did not prevent gets added as a test case in the AP unit tests that all have to pass. Also, each time a human wrecks, it could/will send the data to Tesla to continue simulations and ensure AP would not wreck in the same conditions. How many of the wrecks, even fatal ones, essentially a different driver making the same mistake over and over? Likely millions. Real fleet learning would simply solve that problem once, and prevent thousands of fatalities as a result. As more and more AP capable cars hit the road, this simply will exponentially increase. Thirty years is laughable seeing how much computing has improved in merely the past 20. Maybe 5 tops assuming the legal hurdles are actually surmountable.
- imh 10y ago
  That's such an inethical way to do it. You're suggesting that we just let it learn from wrecks in the real world, instead of trying to prevent those wrecks by other means? This, the Tesla strategy, feels despicable.
  - aramadia 10y ago
    *unethical by the way. I figure the idea is to learn something as opposed to learning nothing like we do today.
  - SEJeff 10y ago
    No it isn't, in fact it is how Tesla AP2 works literally today. What do you think Tesla's new "shadow mode" is? http://www.theverge.com/2016/10/19/13341194/tesla-autopilot-shadow-mode-autonomous-regulations http://www.theverge.com/2016/10/19/13341194/tesla-autopilot-... They also have every single human driver wreck captured with AP capable hardware to make fully autonomous vehicles safer. The way they are doing it right now is brilliant. Edit: As Elon has previously said, not releasing something that saves lives is actually unethical. If a single life is saved by AP software that would have died if the vehicle was driven by human meatware, not releasing it is the unethical thing.
  - vidarh 10y ago
    Any attempt to prevent wrecks we do now are based on learning from past real world experience. All fleet learning does is automate that process and apply it to self-driving systems so that the learning is systematic and will help everyone using said self learning system instead of having to wait for new advice and/or new safety mechanisms.

petra 10y ago

For a deep learning machine with LIDAR , there's only small difference between animals, indian traffic, bad drivers, and narrow/unlit/poor roads. It's just a question of more training(altough a recent NHTSA? regulation says the planning stage cannot use black box techniques like DNNs). And once self-driving is a business, i suspect we'll find ways to get said training data for cheap enough and fast enough. For example Tesla/Uber drivers as trainers .

SEJeff 10y ago

You're grotesquely underestimating what fleet learning will do for traffic issues. The way I see it (as a software engineer myself), is that each wreck that the AP did not prevent gets added as a test case in the AP unit tests that all have to pass. Also, each time a human wrecks, it could/will send the data to Tesla to continue simulations and ensure AP would not wreck in the same conditions. How many of the wrecks, even fatal ones, essentially a different driver making the same mistake over and over? Likely millions. Real fleet learning would simply solve that problem once, and prevent thousands of fatalities as a result. As more and more AP capable cars hit the road, this simply will exponentially increase. Thirty years is laughable seeing how much computing has improved in merely the past 20. Maybe 5 tops assuming the legal hurdles are actually surmountable.

imh 10y ago
That's such an inethical way to do it. You're suggesting that we just let it learn from wrecks in the real world, instead of trying to prevent those wrecks by other means? This, the Tesla strategy, feels despicable.
- aramadia 10y ago
  *unethical by the way. I figure the idea is to learn something as opposed to learning nothing like we do today.
- SEJeff 10y ago
  No it isn't, in fact it is how Tesla AP2 works literally today. What do you think Tesla's new "shadow mode" is? http://www.theverge.com/2016/10/19/13341194/tesla-autopilot-shadow-mode-autonomous-regulations http://www.theverge.com/2016/10/19/13341194/tesla-autopilot-... They also have every single human driver wreck captured with AP capable hardware to make fully autonomous vehicles safer. The way they are doing it right now is brilliant. Edit: As Elon has previously said, not releasing something that saves lives is actually unethical. If a single life is saved by AP software that would have died if the vehicle was driven by human meatware, not releasing it is the unethical thing.
- vidarh 10y ago
  Any attempt to prevent wrecks we do now are based on learning from past real world experience. All fleet learning does is automate that process and apply it to self-driving systems so that the learning is systematic and will help everyone using said self learning system instead of having to wait for new advice and/or new safety mechanisms.