- Joined
- Jun 15, 2004
- Messages
- 3,257
That's not a self driving car... I can clearly see the squirrel driving it in the photo.
Follow along with the video below to see how to install our site as a web app on your home screen.
Note: This feature may not be available in some browsers.
Hope it's not that math error causing havoc again, cause I saw that video and that car did not even hesitate.
But that's the problem, isn't it? People do expect these "modern marvels" to be flawless, or nearly so. Unreasonable? To you and me yes, but then the general public has never been particularly reasonable.The point I was making they could test it in a closed environment for 100 years, and when they started road tests EVEN THEN it would probably kill someone. No amount of testing would be able to perfect it, and the drug comparison is bad because drugs having unforeseen side effects that kill people happens pretty regularly. Expecting these things to be flawless is unreasonable.
We expect a self driving car to perform as well as other self driving cars. It is supposed to "see" obstacles in the road no matter what that obstacle might be.But that's the problem, isn't it? People do expect these "modern marvels" to be flawless, or nearly so. Unreasonable? To you and me yes, but then the general public has never been particularly reasonable.
Ok, I'll bite. I'll go with NASA, as I have pretty good knowledge of that area.
We lost Apollo 1, Challenger, and Columbia. Almost lost Apollo 13 as well.
In every case, when something did not meet expectations, everything was stopped. Before we attempted to put any more human life at risk, the cause was investigated. If it could be corrected, it was, and once it was we resumed operations. It was the responsible thing to do.
Now, if the deaths were directly caused by human error, then there is not much that can be done about it (cars, trains, airliners....). Better education is about the extent.
It is not "pantywaisted" to act in a responsible manner. It is completely irresponsible to not pause and investigate. Fix the problem, if possible, then resume.
"May cause impotence, elevated blood pressure, flatulence and sudden death. Please consult with your physician before taking Adhedra. Adhedra, healing dry skin has never been easier!"
Everything related to those programs was stopped, not absolutely everything was stopped. There is a significant difference there. Ford's self driving program has no relation to Uber's self driving program or Tesla's, GM's, etc, except that they are all trying to accomplish the same thing.
Atlas rockets were being launched within 2 months of the loss of Columbia. Atlas H launched within 2 weeks of the Challenger disaster, on February 9, 1986. Challenger exploded on January 28, 1986.
Yes, they were unmanned missions, but NASA rarely had simultaneous different manned missions happening at the same time. In any case, if the disasters had indicated flawed rockets in other areas, do you think NASA would have risked multi-million dollar rockets and satellites, regardless if whether it was a danger to humans or not?
it would be interesting if the real cause of error, was the nvidia platform and thats why nvidia is suddenly stopping all tests
^and how do we know they aren't?
A pause, autopsy of the event, and restart is expected for UBER, and maybe anyone who uses a significant number of similar component subsystems - But I see a whole lot of "we shouldn't have ANY of these on the roads until we are SURE" in this thread, and that's baloney. We already accept a pretty high number of deaths for our convenience of being able to drive faster than 20 miles an hour. Why is ONE death cause for "remove everything until we're SURE" - well, we'll NEVER be sure. And no amount of in-house testing/simulations will match 100% of real world conditions. Ask any developer, designer, engineer.
BB
But that's the problem, isn't it? People do expect these "modern marvels" to be flawless, or nearly so. Unreasonable? To you and me yes, but then the general public has never been particularly reasonable.
First of all, I have to ask, how do you know all the systems are using different software and hardware? I know for a fact the LIDAR system being used by Uber, is also used by Volvo, Ford and a few others. Same company provided the software for that LIDAR system.
You should find that to be true for most systems. The subsystems provided by third party suppliers will also be providing software. Now, the buyer can certainly modify that software and hardware, but I have not been able to find to wha extent that is true.
To the NASA comparison. If the rocket had a design flaw and was being used for all launches, then NASA would have shut down all the launches. Nothing launched after the accidents had anything to do with the vehicles involved in the accident.
We do not know (if you do, please provide a source as I am very interested in following this) how much or to what extent anything is being shared between any given automotive company. Bosch makes most of the sensors used and they also provide baseline code to the buyer.
Until we know what caused the accident, it would be responsible to shut it all down until the investigation is complete, the cause determined, and a solution is provided. No one is going to get hurt by suspending autonomous testing. It is the in the best interests of everyone to insure the safety of all.
It would be unreasonable to think there would be no delays in deploying this technology Every high risk endeavor suffers a delay, from time to time. It is not unreasonable to delay these programs until the investigation yields the reason for the failure.
They might use some of the same sensors, but Ford might be using a different combination of sensors, i.e. LIDAR in addition to infrared, ultrasound, whatever. Also, how the software reads and reacts to sensor input is basically proprietary to each company unless it is a collaboration.
There is no "might" to it. There are a handful of companies making all the sensors, for the entire industry.
All the companies are being very tight lipped about the exact implementation they are using. That bothers me a bit.
How do you know all the various systems are proprietary? Or is that speculation?
Another thing bothering me is the lack of a standard licensing protocol where the autonomous vehicle has to pass a series of tests before a company can deploy them. There needs to be a standard test suite.
I missed that thread. All I have been able to find on the Net, so far, is mostly speculation with nothing from the manufacturer, until today about the Uber unit.
From what I can gather only two different manufacturers are being used between Waymo, Ford, GM, Uber, and Volvo. Still digging though.