Tesla (and Comma.ai ?) seem to be the only groups that find Lidar unnecessary for L5 self-driving.
There is another group who find Lidar unnecessary for driving: humans.
There is another group who find Lidar unnecessary for driving: humans.
Then there's the part about the brain being what gets the sensor data and interprets it. I doubt Tesla's computer has all the capabilities of a human brain.
Driving behind a boat trailer, it's clear to me it's a boat, but Tesla's computer is confused. Bicycles can be a little iffy as well. In general the way the car perceives the world seems subpar to how a human does, by far.
I'm a fan of the car, I'm mostly a fan of Tesla/Elon, but I don't think these cars with this technology are ever gonna be truly autonomous so they can drive with no human intervention... IMHO.
It could very well theoretically be a software issuse.
While there are technical solutions possible with more hardware, it’s not clear that Tesla could correct for some of these things in just software...
And if the car camera is blinded by the sun, that raises questions about whether the car could safely pull over to the side of the road. I suppose with advanced simulation the car could perhaps predict glare from the sun and pull over ahead of time (“it’s approaching sundown and we’re driving west, you need to take over in 5 minutes”)
For example if the car in front did not even slow down while going through some obstacle (the proverbial plastic bag), then it's very likely that we don't have to either. But if something looks like a drunk guy trying to cross the highway, then it's caution time.
That said, without the actual data/studies/access on what Tesla does and how, it's hard to say how good they are at this. (Sure, in theory everything needed to drive a car well can be inferred from regular human vision visuals. But that means their model has to be able to do counterfactual reasoning - I see this thing as a wall, but other cars go through it, so it's not a wall, so maybe it's a marking. Otherwise it'll very hard to just figure out things frame-by-frame and simply by raw object detection.)
We are very, very far from that. And having such a developed visual processing system is a prerequisite for feeding an executive processing system that can do counterfactual reasoning.
That's also just a shot from the hip, weasely sort of argument but it's harder to dismiss outright.
A camera with those features would be around 3000-15000$, and you'd need two.
Also, the human brain can use focus information and stereo phase to deduce 3D structure, which is another ace up its sleeve.
It might very well be the case that reproducing this is more expensive in training, processing and material than LiDAR. In fact, I'd say it's very likely - we're talking loops that need to update hundreds of times a second and data rates in the gigabits, which our brains deal with by doing a lot of processing inside the eye and along the way.
Current AI tech can do this too, from multiple images. Karpathy even talked about how they do use that in Teslas.
Tesla only does the first.
But just to give an example that is very relevant to self-driving, getting a camera with similar low-light video performance as the human eye costs over 2000$.
So this argument of humans don't need it, is sort of, well yes. We're trying to do better than humans.
So an advanced enough AI could easily be better at humans than driving w/o LIDAR, simply by driving as well as humans when not distracted, and not getting distracted.
It's not that he hasn't invited a little bit of controversy (eg. comments about the diver, covid, etc.) but I think those are inconsequential when it comes to HN's attitude toward him.
The comment regarding humans not needing LiDAR misses the point, which is that either every other self-driving car group is incorrectly deploying an expensive hardware solution or there are valid technical reasons why that might be a better path to L5. Tesla trying to overcome the limitations other companies are deploying LiDAR for is a technical risk they are taking on, and, as the top level post noted, might not be the worst bet given the downside which is that they’ll just have to license the tech.
Implementing L5 without Lidar must be possible. Humans are absolute proof of that. That's not to say Lidar isn't useful, or cost effective given the complexity of the problem, or even just a really good idea that will make autonomous cars better drivers than people. It might be all of those things. All my comment says is that Lidar is not a requirement for driving, and we have a very obvious proof of that.
Anyone saying that Lidar is necessary for driving is ignoring the hundreds of millions of examples of driving with just a pair of 2d cameras and a highly specialised organic supercomputer.
EDIT: The point another poster made about weather and bright sunshine is a reasonable one - humans can't drive when they can't see either, so Lidar would potentially make an autonomous car better at driving than people. That would be an example of how Lidar is useful, but still not necessary.
Humans also have a organic supercomputer in their brain. Variety of sensors is probably the best approach. I've grown up and seen technology prices on all sorts of weird technology fall (3d printers, sensors, midi controllers). At some point, we will basically give up on perfection on LIDAR and just do cheap multi-LIDAR arrays where the error rate is overcome by just having multiple units.
We can say that something better than human-level driving ought to be possible without LIDAR, though; take the perception of a human (cameras) and replace the brain with a computer that can make better calculations much faster than a human can.
Given that Tesla are the only company to have put something approaching autonomy into the hands of customers, that does seem to be the case.
What happens next in this thread? Someone points out that Tesla's solution is still far from perfect and claims that they're being highly irresponsible by having shipped it. Then I'd ask how many accidents have happened as a result of FSD beta having shipped to some customers already and they assert that Tesla isn't transparent with the data so we don't know. I'd point out that if we don't know of a single Tesla FSD accident so far, it can't be that irresponsible to have put it in the hands of a few customers and that it's been rapidly improving in the weeks since. It's the same thread, repeated over and over on here, Reddit, etc.
Tesla was the company that made electric vehicles successful. To your point about groups taking different approaches, I think Musk & Tesla do take approaches that will differ from others, and succeed partly because of their approach being better in some important way. I think they had a bit of an unfair advantage when it came to autonomy. They needed hardware that could be shipped on vehicles as soon as possible, which meant that it needed to be cost efficient & energy efficient. LIDAR's energy efficiency doesn't get talked about much, and I don't mean just the LIDAR itself but also the subsequent processing power required to make sense of the resulting point cloud. Each Waymo vehicle is an expensive tech demo. It's a very good tech demo, but over the years it's become evident that they're very hesitant to scale it up right now. I'm sure they'll eventually get there, but they are not a company focused on getting this done ASAP. Tesla are getting autonomy done with urgency.
Well, that’s Waymo (in Phoenix) which has put out anything approaching autonomy, not Tesla. Tesla requires a driver.
Tesla itself uses HD maps for things like intersections, traffic lights and so on. It's just not at the level of detail others use it for.
They’re also offering a commercial service. Which means figuring out operations, customer service, emergency protocols, working with local administration and so on. They’ve said a larger goal in Phoenix is to figure out how to perform and scale operations. The local regulations there are much more favorable. Until recently, Phoenix was one of the only (probably only) places where you could operate a robotaxi program and charge for it. California just recently approved it, so SF is where they’ll likely go next as they are already setting up everything there.
Ultimately, mapping is the not the constraint for them to go to new places. It’s the operations and regulations that they need to figure out every time.
They also have lidar solutions, but they certainly haven't ruled out camera only.
The issue people take is the reckless approach to selling nonexistent or immature tech that by its nature puts human life at risk. It is possible to both respect and disagree.
The arguments used against Tesla are in fact legitimate if we’re talking about a regular company. But Musk has proven to do the “impossible” multiple times and has been right on so many levels, so at this point you have to question whether he’s catastrophically wrong about this, or perhaps the others are about lidar.
Either way, the confidence that Tesla is wrong is way too high in my opinion.
People need to stop fantasizing about 'Elon' being some great guy because they like what he achieved. It's almost like the technorati need their own version of celebrities to fawn over.
I for one am not remotely jealous of him but disappointed that others think his behaviour should be emulated or worshipped just because he has been successful.
I'd rather see someones imperfections and know where I stand than have another generational #metoo
Both unofficially and when law is broken.
Or maybe unenthusiastic about the fact that all he has is a vision, but has never built anything by himself. When you get kicked out because _Peter Thiel_ is more tolerable than you, that says a tiny bit about his character.
It's not "a little controversy". It's being a sack of shit of a human being.
the only thing he is good at is marketing himself
people think he is innovative because he sold decades old tech to people who didnt pay attention before, because he is a glorified marketing exec and hes not even good at that
consumer satisfaction is a poor way to measure quality