If he’s true to his word. I wonder if Ford can close the gap this model cycle with their program due next year
This is a rewrite of existing functions to take advantage of Tesla’s new processors—moving away from the nVidia processors. It should be faster and help in the AI’s decision making.Is 6-10 weeks Elon speak for January?
Tech advances are good in the industry overall. But I won’t be buying a Tesla anytime soon, no matter the news.Tesla Battery Day is now combined with Tesla Stock holder meeting 22 September. This might be the day he releases more info or teases the world. After recent Space X successes everyone is looking for something big from Tesla too.
Despite what Elon claims, I don't think Tesla itself has ever said that its fool proof in all situations. All the warnings on the screen and so forth say you are the one in control, not the car. For the crashes you cite I still blame the stupid drivers, not the car.We have seen the ridiculos erors that Tesla software has made, over and over agian : running into huge fire trucks parked in the highway, brick walls. highway concrete dividers, etc. The landscape encompassing places and situations that a car can encounter cannot ever be enumerated by the software engineer, much less the appropriate response/. Autonomous driving is a pipe dream.
"Almost at zero interventions on my drive..."
I agree with most everything you said. But I don't know if they have to be 99.9% accurate. There are self-driving semi's between Tucson and Phoenix. They don't need to handle the snow situation, or someone jumping out in front of the car (on the freeway at speed in a semi, your screwed no matter who is driving)."Almost at zero interventions on my drive..."
Think about that statement for a bit. For any sort of autonomous system to work, you would have to have zero interventions for maybe a hundred million miles. Consistently.
If Elon is admitting that the present system is so bad that it requires interventions on short drives, it isn't anywhere close to handling all of the edge cases required for full autonomy. It is easy to get to 90 or even 95%. It is a thousand (ten miillon?) times harder to get from 95 to 99.9%. Because that is where the oddness exists, and human brains can handle that oddness.
Don't get me wrong. I want autonomous, and I also thought it would happen sooner, but every few thousand miles when I'm driving, something very odd happens. I handle it, shrug, and go on. But I don't know how a computer would do. And an accident every few thousand miles is not the standard. Not even close.
Humans have a fatal accident approximately once every 86,000,000 miles driven. You read that correctly: as bad as we are, distracted, using cell phones when we shouldn't, fatigued, drunk, etc., we only kill a human once every 86 million miles.
We will have autonomy and it will be great. But it is a very hard problem.