He says it's better than anything other than the Tesla, yet he ran away the moment NHTSA looked into his project. I've been around long enough to know when something smells like bs, looks like bs, and sounds like bs...it's bs.
If there is going to be a new AI "winter" (and I'm starting to think that's more and more likely every day), it will probably be the result of too much hype around overblown ego-driven claims such as this. Hotz is probably doing more harm than good at this point.
Do you think the media can be overinflating the whole project?
Furthermore, ACC and LKAS are standard features provided by Honda/Acura with the required feature package needed to run this code! Neo will provide more functionality, but as of now, NeoOS hasn't been open sourced at all and is just a binary Android image.
It's not just the media though, he has a history of drama and overplaying things and is repeating the pattern here. I don't want to downplay his achievements, he's done a lot, but the hype is far greater than even that high bar of accomplishments.
I'm seeing a lot of internet commenters reacting emotionally to Comma.ai, but the only thing they're getting upset about is their own inflated expectations. The actual product is too niche, too esoteric to warrant a strong reaction. It's a toy for nerds to tinker with, nothing more.
> It’s about on par with Tesla’s Autopilot 7, the one that came out when it launched.
That's not playing it up to you?
He's using a car that already has ACC and LKAS functionality to provide exactly that, yet scorns Tesla for using Mobileeye.
That part... ;)
Let computers handle the boring parts of driving, and send the hard stuff over to the human when it needs to.
That is a lot of gains that we could be running right NOW on the streets.
If the human driver mentally switches off or stops paying attention, things get dangerous as has been demonstrated in numerous Tesla autopilot crashes.
That isn't to say that these features don't improve safety. Widespread introduction of radar-based automatic emergency braking (AEB) will significantly reduce crashes and save many lives. But until we're closer to 100% autonomy, features that reduce driver workload need to be implemented carefully.
I'd be happy if this type of system would scan really aggressively to warn/avoid the deer barely visible off the road before it jumps out in front of me...
edit: grammar corrections
If rules, guidelines, and heuristics can beat neural nets at driving (which they might, I agree with you on that) it won't start a new AI winter because it won't erase the other things that neural nets are good at (e.g. classification).
In my opinion, real advances apart, a large portion of the "new AI bussines" is just a way of getting money from investors telling following magic words: "deep learning", "big data", "jump the wagon before it gets too late", etc.
What hard work? It's a COTS part.
Not to mention, front-facing radar, even combined with image recognition is woefully inadequate for true self-driving. (As opposed to, for example, Tesla's so called "Auto Pilot".
Cameras + radar is more than humans have at least, so I think that should be enough.
Solid state LIDARs will soon be cheap enough and a good complement of course but I find it hard to believe autonomous driving would be impossible without it.
Imagine two companies competing to win at Go, and one company had the attitude that megawatts of energy was not necessary, and another company threw whatever resources they could. The second company just played top-level Go this year. The first company is at least another 5-10 years away from it.
I'd also like to point out that humans do SLAM already (LIDAR solves SLAM). From a previous comment:
Humans implicitly perform SLAM (simulataneous localization and mapping). What do I mean? Look around your room. Close your eyes. Visualize the room. As a human, you've built a rough 3D model of the room. And if you keep your eyes open and walk through the room, that map is pretty fine-grained/detailed too and humans can keep track of where they are in the map.
Maybe we are just arguing over terminology.
Firstly, the state of the art (which is not deep learning based) still has trouble when the camera doesn't turn/swivel a lot (e.g. it has trouble when the camera is just moving straight).
And that's not even getting to the most critical assumption. All of visual SLAM right now assumes the environment is static (!!!), and doesn't work when things in the environment are allowed to move.
execpting perhaps, that kilogram lump of gray matter floating around in our skulls.
My guess is sometimes it's probably ok to decide a project is not yet in a position (financially or structurally) to take on the additional responsibility of dealing with an enormous bureaucracy such as the US govt. And if that means you have to give your product away in order to do so, that's a calculated risk you might feel you have to take.
I seriously doubt that when NHTSA contacted them they were like, "just invite us into your building and show us around", like a health code inspector might.
As for the money
>Well, we have Andreesen Horowitz [the VC firm], which provided $3.1 million, and we spent under $1 million to do all of this.
OTOH, this still seems a bit like a hobby project to me. If I can believe the other comments, the code is not too spectacular, and the biggest asset is likely Hotz' name.
Or the funding is just to buy the person/people/team to make some innovative products or work in some other self driving car research teams?
But when nobody is buying comma.ai, nothing of that will happen. The (little, given the domain) money invested will only have been sponsorship of a romanticized take on hacker culture. Which may actually have been a contributing factor in the investment decision here, this a frugal garage operation, not a fast-burning startup with fancy offices.
Open sourcing some code can create visibility, without it we would not be talking about this today. From the investment perspective, opening this code is clearly a hail-mary-pass. There are situations when those are the best course of action and I think this is one of them.
Do you see anything wrong with a hobby project steering cars around at high speed? Hotz is clearly a very clever guy, but his judgement is the issue here.
To be fair, this could have nothing to do with the quality of the project. I don't want anything to do with the NHTSA (or any other government regulatory body) if I can help it.
Great, so don't make software for self-driving cars.
To be fair, looking at the code, it has all to do with the quality of the project
The rest is basically just a bunch of glue and bootloader code.
Here there is a discussion about the github repo
Sounds like your opinion of him is even worse than the above.
http://www.embedded.com/electronics-news/4423365/Toyota-Camr... http://www.edn.com/design/automotive/4423428/Toyota-s-killer...
They are basically looking for proof that you did required things such as specifications, tests, risks management... They are looking into your development process to estimate if your product is safe or not. But this is only indirect.
Until you actually kill someone, and then, the same agency will send you its investigators and auditors.
UL actually do testing, although they're not exactly a regulatory agency.