Apple, Tesla Ask California To Change Its Proposed Policies On Self-Driving Car Testing (reuters.com)
Tesla and Apple have asked the state of California to change its proposed policies on self-driving cars to allow companies to test vehicles without traditional steering wheels and controls or human back-up drivers, among other things. Reuters reports: In a letter made public Friday, Apple made a series of suggested changes to the policy that is under development and said it looks forward to working with California and others "so that rapid technology development may be realized while ensuring the safety of the traveling public." Waymo, the self-driving car unit of Google parent company Alphabet Inc, Ford Motor Co, Uber Technologies Inc, Toyota Motor Corp, Tesla Motors Inc and others also filed comments suggesting changes. Apple said California should revise how companies report self-driving system "disengagements." California currently requires companies to report how many times the self-driving system was deactivated and control handed back to humans because of a system failure or a traffic, weather or road situation that required human intervention. Apple said California's rules for development vehicles used only in testing could "restrict both the design and equipment that can be used in test vehicles." Tesla said California should not bar testing of autonomous vehicles that are 10,000 pounds (4,535 kg) or more. Tesla also said California should not prohibit the sale of non-self-driving vehicles previously used for autonomous vehicle testing.
To fuck you in the butt
Breaker, breaker, that'd be a big no on ditching the steering wheels, back up drivers, and reporting fack-ups, good buddy, over.
Fact is, autonomous driving systems aren't yet up to snuff to go the full monty. Until they prove out, they need a human with some skin in the game, and who's aware s/he's playing. And the state can't be sure how close to the tipping point we are without reporting.
On the flip side, I agree that allowing higher gross vehicle weights should be allowed, the better to test freight hauling.
Luke, help me take this mask off
Disclaimer: This is just my common sense talking; I feel like there has probably already been extensive testing done in closed-course situations and computer simulations. The problem with an emerging technology of this type is that you cannot account for the human element without a great deal of exposure to it. That's for both the programmers of the car's software, and the other drivers on the road. What your stating sounds to me like wanting a toddler to run before it can walk when it just learned to crawl a few months ago. Apple/Tesla/Alphabet et. al. are not asking for the go ahead to sell self-driving cars, they just want the regulations to be more friendly for their side. That being said, I feel you are right that there should still be extensive testing (though I think it probably qualifies for on-road at this point), and just as important; I feel as though it should be as transparent as possible without the divulgement of trade-secrets. I'm a realist, I understand that everyone's in it for a buck. Side note: Reading your comment took me a few tries when I stumbled over the "blind folded" part. I kept thinking you meant actual blind drivers folded in half.
As someone that has actually provided a learning experience to a self-driving car (I spend far too much time walking downtown Mountain View streets), I can say with some confidence that self-driving cars are far more careful to avoid incidents than a human driver would be under the same circumstances. I trust them far more than most of the drivers on the same streets (and, again, I see plenty of them, and have observed them in real time). And at least they are not looking down at their phone driving through the stop sign as I am crossing the street. I will never assert they are perfect, but perfect is the enemy of the good enough, and they are far better than good enough (and certainly than the average SV driver) today.
There are ways to safely test complicated scenarios with multiple cars. Look at how the highway safety institute handles crash tests. They make a car and essentially destroy it. That is how you safely test cars by having several drivers on a closed course simulate heavy traffic changing lanes and even tapping the vehicle to simulate a collision.
It very much is possible to make them go totally without a driver. Not from a legal standpoint, but from technical standpoint no problemo, the car will drive on its own if told to do so. Safety of such a thing is questionable obviously, but how are we ever going to answer that question without experimentation? Is "it worked fine on test track" good enough answer to you? Its not sufficient imho, only real life statistics can give a satisfactory answer. Traffic accidents are sort of an acceptable loss in today's society already, they are a fact of life we must accept anyway, so i think its not that big of a leap to accept a risk of testing driverless cars in the name of lowering traffic causalities in the future. Mind you the driverless cars being tested are probably already safer than human drivers or so the developers claim at least.
There are ways to safely test complicated scenarios with multiple cars. Look at how the highway safety institute handles crash tests. They make a car and essentially destroy it. That is how you safely test cars by having several drivers on a closed course simulate heavy traffic changing lanes and even tapping the vehicle to simulate a collision.
Yes, you can create complicated scenarios and test them, but you can't create nearly enough to cover the range of what will actually occur in actual driving conditions and interactions with human drivers.
How many years did you spend on a test track before they let you on our public roads?
E.g., the Google car has already encountered a jay-walking wheelchair user chasing a ... was it a duck?
When would you even THINK to simulate that?
I think we've pushed this "anyone can grow up to be president" thing too far.
Sorry, but it's a reasonable request. It may also be reasonable to deny it, but it's a reasonable request. There's no way that a "driver" who's just been sitting there playing a game on his phone will be able to take over the driving in 10 seconds, so the steering wheel is useless in emergencies.
I think we've pushed this "anyone can grow up to be president" thing too far.
Except the test tracks are designed to test the mechanal features. For automated cars there is a degree of this on test tracks. But there comes a point where it needs real world testing. If you write software at nearly every level of complexity once you hand it to the real world they find new problems that needs to be fixed.
If something is so important that you feel the need to post it on the internet... It probably isn't that important.