this post was submitted on 11 Oct 2024
25 points (61.1% liked)
Technology
58669 readers
3897 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
if he gets over his fear of lidar and stops trying to do everything with cameras this could actually work out ... so basically this isn't going to work out.
No it won't, Elon Next year Musk has promised Fully autonomous driving next year since 2016. He has sold his cars with subscriptions for FSD for years, despite it doesn't work. It's even illegal to call it FSD now, so Tesla has to call it assisted FSD, which is an oxymoron.
With this move, Enron Musk will more likely ensure the continued decline of Tesla. The Cybercab most likely will not be a moneymaker, and the focus on developing it, will detract from Tesla developing much needed new EV models, for a market with increasing competition.
Tesla is far from #1 in developing autonomous driving, so the chances are very slim that they would be even close to be first to market.
AFAIK this is pretty much the current rank:
-1 Waymo (Google)
-2 Mercedes
-3 Mobil Eye
-4 GM (Cruise)
-5 Baidu
-6 Tesla
Possibly Nissan-Renault (WeRide) and Nvidia can match Tesla too.
Notice that Tesla used MobilEye originally up to 2016, but MobilEye ended the partnership after a Tesla model S had a fatal crash. I suspect the irresponsible claims implementation and practices by Elon Musk were too much.
Elon Musk is insane and a con man, to believe anything he claims about the future of his companies is naive.
Because Musk has a weird obsession that everything needs to work with just cameras, and no other sensors can help. While it might work someday in the far future with proper AGI (e.g Delamain from Cyberpunk 2077), until then it's a pretty hopeless endeavour.
The problem is computer vision has a LONG way to go before it’s truly on par with human eyesight. Musk loves to crow how cameras are sufficient since we use our eyes to drive.
The thing is, eyes have special neural circuits that detect motion. They essentially filter out unnecessary information and send just the motion details to the brain. This prevents the brain from being overloaded with every detail the eye constantly sees.
And being overloaded with everything is exactly what computer vision currently does. It’s just a stream of images that the computer must analyze completely. So it’s working exactly opposite to how the eye & brain works.
Long way, but it's not an impossible task, as at the core the eye is nothing but a bunch of light sensors that spit out a result, we just need to figure out how to calculate that result ourselves. Motion amplification could be one solution given enough computing power to do it in real time, for example.
But we agree, safe and accurate camera based self driving isn't going to happen in a long, long time.
I'm not sure the approach Elon Musk has to developing self driving will ever work.
From what I've heard about how they "teach" the AI, it probably won't, because loading massive amounts of new data is rewarded, but there is no proper qualitative control.
Man, how did Mobil Eye not just tack an “s” on the end? The pun is right there, and it’s frankly excellent and super topical to what they’re trying to do.
What benefit would a lidar bring that they haven't already achieved with cameras and radar? The car not seeing where it's going is not exactly an issue they're having with FSD.
a lidar could tell the difference between a person on a bus billboard and a person. it brings 3d to a 2d party.
A lidar alone can't do that. It'll just build a 3D point cloud. You still need software to detect the individual objects in there and that's easier said than done. So far Tesla seems to be achieving this just fine by using cameras alone. Human eyes can tell the difference between an actual person and a picture of a person too. I don't see how this is supposed to be somethin you can't do with just cameras.
Human eyes can't do it alone either. The brain has to process the information delivered.
Funny, last I heard, Tesla FSD has a tendency to run into motorcycles.
With lidar there would be no doubt that there is an actual object, and obviously you don't drive into it.
are the tesla cameras 3D?
No, and neither are your eyes, but you can still see the world in 3D.
You can use normal cameras to create 3D images by placing two cameras next to each other and creating a stereogram. Alternatively, you can do this with just one camera by taking a photo, moving it slightly, and then taking another photo - exactly what cameras in a moving vehicle are doing all the time. Objects closer to the camera move differently than the background. If you have a billboard with a person on it, the background in that picture moves differently relative to the person than the background behind an actual person would.
Talk about making a difficult problem (self-driving) more difficult to solve by solving another hard problem.
Just slapping on a lidar doesn't simply solve that issue for you either. Making out individual objects from the point cloud data is equally difficult plus you're then having to deal with cameras too because Waymo has both. I don't see how you imagine having Lidar and cameras would be easier to deal with than just cameras.
Also. Tesla already has more or less solved this issue. FSD works just fine with cameras only and new HW4 models have radar too.
That's a grossly misleading statement.
We definitely use 2 eyes to achieve a 3D image with depth perception.
So the question is obviously whether Tesla does the same with their Camera AI for FSD.
IDK if they do, but if they do, they apparently do it poorly. Because FSD has a history of driving into things that are obviously (for a human) in front of it.