Tesla says its vehicles that are equipped with the latest versions of its vaunted “Full Self-Driving” system can travel from point to point with little human intervention.
Tesla keeps promising things are fixed in the next update, then the next update, and so on. I don't think Tealas have the proper sensors to avoid collisions and their algorithms don't think like an attentive driver does.
The company has cautioned that cars equipped with the system cannot actually drive themselves and that motorists must be ready at all times to intervene if necessary.
This describes a level 2 system...
And in less than two months, the company is scheduled to unveil a vehicle built expressly to be a robotaxi.
...but this would require a level 4 system.
“It’s not even close, and it’s not going to be next year,” said Michael Brooks, executive director of the Center for Auto Safety.
This will be the reason Tesla falls behind the rest of the automotive industry, wasting money on vanity projects instead of developing better vehicles.
The 3, Y, and the huge number of Chinese EVs being sold around the world have shown there is a huge market for affordable, practical electric vehicles, and what are they developing? A vehicle that won't be able to fulfill it's intended role for a decade almost everywhere.
I think the focus on AI is what will be the problem. Sure, AI is cool, and sure you need advances for self-driving, but you’re a car manufacturer and can’t neglect car manufacturing
Aren't level 4 systems still illegal in the US anyway? I remember Volkswagen holding off on a minibus due to this limitation when they managed to create a working proof of concept 8ish years ago
My understanding is that they're not illegal in so much as they would have to be proven to actually work. Since no one's ever been able to do this to the satisfaction of the regulators effectively self-driving cars are illegal.
We have Tesla's on the road anyway, so I don't quite understand how that works.
They are only effectively illegal right up until they're not. If a company came up with a genuinely self-driving vehicle my understanding is that it would be authorized but they would have to actually demonstrate it, and that's possibly more than the corporations really want to take on at the moment.
Presumably being second to market is the more cost-effective option so everyone's holding off until someone does it first.
Also this seems to very much be them suffering from the 90/10 problem that some many software projects suffer from: you can do 90% of the work with 10% of the effort but that also means it takes 90% of the work to do the other 10%.
Earlier people from the outside perceived Tesla as being ahead of everybody else on this because they just started before just about everybody else, were riding the easy part of the curve were you're constant delivering improvements because you're just tackling the easy bits, and they were probably even doing a rush job and taking shortcuts to maintain that apperarence of being ahead of the rest because that maintained an appeareance of Tesla as a Tech Stock rather than an Auto Industry Stock, hence with market valuations 10x or 20x higher than they would others which meant sky-high Tesla Stock Prices justified by how "technologically ahead" they were in a "key future technology" - the company was just executing a typical conman strategy of looking like they were making it in the hope that their early-mover status and the increasing investor funding that strategy pulls in would allow them to actually make it before everybody else or at least with a larger installed base, similarly to how Theranos was doing only unlike that company Tesla did just the right balance of deceit and reality to just be on the right side of the Laws for Fraud.
Tech companies absolutelly can get away with doing this during the early and easy parts of the project because for non-experts it looks like they're doing fast progress - which is why Startups nowadays (which is an Era of way more bullshitty and even fraud in the Industry than, say, pre-2000) commonly do things exactly like this - but then they reach the hard part, progress speed naturaly goes down a lot (the project transitions from the 90%-results/10%-work speed to the 10/90% one) plus all the the early shortcuts (a.k.a. Technical Debt) come due to be paid for: "out of control car careening down the hill" meet "concrete wall".
Finally, I wouldn't at all be surprised if they're stuck down a dead-end for the technology which is the wrong way to reach level 4 and have to go back and redo much if not most of that work they did as a rush job to keep impressing investors and ill-informed customers with how "ahead of the pack" the were. Under the leadership of Musk I suspect that they will never be able to reach level 4.
Tesla's Full Self Driving does not exist. If it did, it would be used in the ridiculous loop they built in Vegas which is literally like 10% of the complexity of driving in a real road... yet it is not even good enough to be piloted there.
Tesla held a commanding lead over the other automakers in the self-driving segment a few years ago. Now they’ve all mostly caught up thanks to Musk’s unhinged firings. Tesla lost some of its best talent for no other reason than not wanting to work for an egomaniacal billionaire nut job.
Tesla needs to fire Musk before he runs it into the ground just like he’s done to Twitter.
They're only ahead of everyone else because they were prepared to release a product that was untested and quite possibly unsafe, whereas the other car manufacturers realized that would be detrimental to their business, both in terms of reputation, and the inevitable lawsuits. Tesla just does whatever though.
other car manufacturers realized that would be detrimental to their business
Ummm do they? I don't care enough about cars to remember the details, but I'm sure I've heard some controversies about safety for at least a couple of brands
Because they weren't willing to open themselves up to the lawsuits for rushing a half baked autonomous driving function. Their systems likely work just as well as Tesla's, which is why they wont advertise it as full self driving because it kills people.
Humans are not safe. 40k of them get killed in vehicle accidents every year in the US alone. Self driving doesn't need to be flawless. It only needs to be safer. If perfection is the only thing we'll settle for then that'll cost us hundreds of thousands, if not millions of more lives untill we get there.
If we replaced every vehicle on US roads with self driving cars that were twice as safe of an driver as average human is, there would still be 50 deaths every single day. That's 50 daily news articles on Lemmy about how "not safe" self driving cars are despite them saving additional 20k lives every year.
Honestly there's so much shit about Tesla's software on the Internet. I own a model y with "FSD" and I can confidently say that the current version (12.5.3) is a 90% solution. Is it perfect? No. Can it do everything? No. Can it drive me to work and back, to the grocery store, and whatever else I need without me intervening? Yes it can, and has no issues 90% of the time. Obviously for it to be considered level 5 or whatever it needs to work 99.999% of the time, but it's good enough right now for me to not only use it regularly, but to also enjoy using it.
It's an awesome piece of software and it still blows my mind to watch my car drive me around. We are living in the future.
I recently rode in a Tesla on FSD for over 26 hours of freeway travel. It was flawless the entire time.
On city streets? 90% was about right. It once took too sharp of a turn at a double right turn and spooked the driver next to us (although it didn't cross into their lane, just got close), and another time decided to only change lanes halfway into the left turn lane.
I agree that it needs to be near 100% on city streets before it's ready for launch because that 10% difference is HUGE when it comes to safety. If their "level 5" taxi isn't using some vastly improved software, it needs to be kept off the street.
To every even slightly educated person on the matter it's immediately obvious that the majority of people commenting on these threads don't have up-to-date information about how far FSD has come and their opinions are based on how bad it used to be and / or how bad they wish it was.
Everyone is free to go to YouTube and watch videos of people intentionally pushing the limits of this system. Like you said; it's not flawless, but damn it's good.
There was an story a while back where it was discovered that Tesla focuses their efforts on the routes taken by creators (and other influencial people), to intentionally make self driving look better than it is.
I have it and it is legitimately a great driver aid. But you need to understand its limitations to use it safely. It's a really awkward place to live in, because I get an immense amount of value out of it, but I would not just throw my mother in the car and trust her to use it safely.
I'm considering buying one but lately I've been thinking about maybe getting a Mustang because the BlueCruise seems more honest. However, Ford doesn't seem super committed to electric and it's all hard to justify when my Honda just won't die.