-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Autonomy Investor Day - April 22 at 2pm ET
- Thread starter diplomat33
- Start date
Now this is more interesting, the triggers evolving. Too bad Elon had to interject with some mumbo jumbo about ”every mile” training networks, which is clearly not happening (nor probably will ever happen) — and the network training still happens in the back end anyway — but the cut-in trigger certainly is interesting.
My 12 year old loves Alexander's talks. He is waiting to get home to see this today. Just really energetic about learning.
Last edited:
Elon’s jokes just diminish this, he should just get off stage for this. The presentation is very interesting itself.
We have known about the triggers that are running there for a very long time, there was a list of them even. But this is a good update on what they are doing today with them.
Shadow mode itself is a bit of a marketing term though. It is trigger-based, not constant.
Not that big of an issue from an investor standpoint. Future looking. Sure, they have some risk with those of us with AP2.0 cars out there... but the future is what really matters here (IMHO).
I don’t think there is much risk for us AP2.0 owners specifically. I don’t understand why people think there is risk on that one.
I am not convinced by their reasoning against Lidar. Even Andrej uncharacteristically disappointed there.
Sure if you have poor vision then Lidar can be a crutch but others have much greater vision than Tesla has and have Lidar in addition to that... so the argument does not stand... it felt like something Andrej was made to add there. Quite uncharacteristic considering they showed how other redundant sensors in Tesla were a benefit... suddenly Lidar only is a hinderance...
I did like the update on the triggers.
Sure if you have poor vision then Lidar can be a crutch but others have much greater vision than Tesla has and have Lidar in addition to that... so the argument does not stand... it felt like something Andrej was made to add there. Quite uncharacteristic considering they showed how other redundant sensors in Tesla were a benefit... suddenly Lidar only is a hinderance...
I did like the update on the triggers.
Ludalicious
Active Member
You would think a class action lawsuit would be in the interest of an investor.
Sure if you have poor vision then Lidar can be a crutch but others have much greater vision than Tesla has and have Lidar in addition to that... so the argument does not stand... it felt like something Andrej was made to add there. Quite uncharacteristic considering they showed how other redundant sensors in Tesla were a benefit... suddenly Lidar only is a hinderance...
Yeah, I can see LIDAR as still being useful as a back up, especially in inclement weather where vision might be slightly impaired.
The questions actually asked if Tesla could train NNs in-car, Musk answered HW3 is not optimized for that. So no change there.
Yes, most importantly it is a redundant sensor with different benefits from vision. Everything they say about training redundant sensors applies to it so it is a bit hard to take the attack on Lidar seriously. Seems something they tacked on to convince investors it really is not needed.
I would have actually felt better without piling on Lidar and instead just convincing us on their vision. It would have felt more believable in my view. ”Lidar and many kinds of sensors can certainly be useful but we feel we have a optimal solution for our great vision engine” and this allows us to get there first etc etc...
Now they are trying to convince us a bit too much on that point. To the point where it is not that believable.
Yes, but how many AP2.0 cars are we actually talking about. I feel like it is small potatoes compared to potential sales I the next 10 years.
(As I mentioned, I do have AP2.0)
Daniel in SD
(supervised)
Yeah, except that if people believe that LIDAR is an absolute requirement for self-driving, then no matter how good your vision is, it won't matter to them. So it is not good enough to prove that you have excellent camera vision, you have to make the case that LIDAR is not needed.
Then that case so far has definitely not been made. I’m not sure badmouthing your competition is ever the best tactic but I guess they feel they can change the industry narrative so they try — instead of just convincing us their solution works great without Lidar and ignoring Lidar...
I’m not. I like the technical presentations. This is a first for Tesla on Autopilot, we’ve heard many from the likes of Waymo and MobilEye of course.
We’ll see how much new progress we see, who knows, but at least we get fresh insight into their development process.
Similar threads
