Hacker News Reader

Comment by noonespecial | Hacker News Reader

There's 2 possible reasons.

1) It really is near vaporware just being shown off in perfect conditions to seem exceptional.

2) It actually does work that much better with only $1000 worth of hardware and responding would give away the secret sauce.

I don't give much credit to the theory that they didn't see this coming and were caught flat-footed by the sudden requirements.

Oh and:Final speculation: there’s a reason it was called the Comma One. Where there's a Comma One, there's likely to be a Comma Two

I was really hoping it was actually a C64 reference, namely: "where the 'drive' goes"... ,8,1

View on HN

Or, you know, the gigantic field of possibilities between those two things: it works a little bit under some conditions.

Fortunately for us and unfortunately for Hotz's ego, regulators look down upon things that work a little bit under some conditions when they also happen to control two ton metal cages capable of traveling at 120mph.

> it's [the] NHTSA that remains behind the curve in evaluating the myriad different technologies

Uhhh... that was exactly what they were requesting to do with their letter.

The one fatality with Autopilot happened because the system worked too well, the driver came to trust it when he shouldn't have. The shittier the Comma one is, the more in-the-loop the human has to be, and risk compensation becomes less of a factor.

What in the fucking hell are you talking about?

A system that is seemingly half blind (repeatedly crashes into vehicles that only partially obstruct the lane), decapitates a guy, crashes another one into a fucking post

It didn't work too well.

It pretended to work too well and gotten people killed. There is a reason MobileEye walked away from Tesla

They are being rather cavalier with their attitude to safety, apparently. (Apparently the falcon wing doors can go nom nom on you if you are not careful too - sure the guys are laughing their asses off, but it's not a good design)

https://www.youtube.com/watch?v=TdPPp3hoGQ8

EDIT: OR, EVEN BETTER: http://jalopnik.com/did-tesla-quietly-remove-a-safety-featur...

One fatality in 200 million miles of driving reflects a cavalier attitude to safety? It's not like the guy who died in that accident wasn't familiar with autopilot's reccommended usage, he read the manual, Tesla made it pretty clear you still gotta watch the road while it's engaged. I mean, if you're into misusing cars, any of them will allow you to accelerate into a brick wall if that's your inclination.

I'm not sure what Falcon wing doors has to do with any of this.

There is at least 1 other death in China where autopilot is cited as the cause, and there are many other accidents in China which are not widely reported in the United States (it's easy to find several on Google though, many of which are even posted on YouTube with video proof).

Statistically speaking, Tesla's autopilot is more dangerous than regular driving. They spun bad statistics to claim otherwise. They ignored the difference between highway driving (which is safer) and all driving, and ignored demographics (most Tesla drivers are older and affluent).

It is too early to really draw many conclusions about autopilot, because there is actually very little data. Comparing the number of miles driven by Tesla to the number of miles driven by all other cars in one year is a difference of several orders of magnitude. But they are under investigation, and as more time passes the more damning it seems for Tesla.

It's true that there's too little data to draw conclusions. A Toyota spokesperson claimed we'd need about 8 billion km of autonomous driving to draw statidtically relevant data. and of course, by the time a single company, or even all companies combined have gathered that many miles the technology will have progressed and the early data will be irrelevant.

Safety arguments are only meaningful in these sorts of abstract discusssions, though. Keep in mind that when we decided to place the automobile at the centre of American life in the early postwar era, fatalities per vehicle mile travelled were 10x what they are today. We didn't do it because it was safe, we did it because cars are cool, and some big companies stood to make a lot of money.

You think maybe part of why there haven't been more numerous malfunctions might be because of rigorous NHSTA testing of various parts of the automobile?

Also just because a person on birth control gets pregnant doesn't mean we start pushing pills with unknown success rates in order to get people to use condoms.

If you're imagining that autopilot is anything more than driver assist technology, then there already is numerous malfunctions, autopilot fucks up regularly. With or without NHTSAs involvement, Autopilot (or Comma one) wasn't designed to be safety critical autonomous driving technology. NHTSA has about as much business regulating it as it does regulating whether or not you can hang dice on your rear view mirror.

If it were a form of birth control, level 2 would be the pull-out method: no guarantees, so be vigilant.

Nope. I'm imagining that NHSTA keeps people safe, and I don't want a car anywhere near me that's controlled by something that the NHSTA hasn't had a chance to even evaluate.

It's foolish to believe that real humans, driving real cars, will treat these technologies as mere "driver-assists." Autopilot is used -- shockingly enough -- as an autopilot. And even if they /were/ used perfectly, these systems are exerting physical control over the car -- something that the National Highway Traffic Safety Administration has a distinct interest in managing. The NHTSA would be grossly negligent if they did not attempt to regulate these devices.

I have yet to see the simplest proof that comma ai work is just a cute trained NN over a camera, piped to basic controls, with some potential creative ideas (in a conf, hotz mentioned working on time based depth inference instead of the traditional stereo camera setup).

I'm biased for conservative companies, I have no idea what they're doing and how much resources. Maybe they're doing something at the same level. I believe they have more refined physics model to compute surroundings and car physics in a way to ensure sane and safe behaviour. Exception made of Tesla AP, which I always found borderline lying.

The medias and the crowd seeems to believe wild hacking and current year tech is giving you better products, but physics are physics. I'm still waiting for a proof of solid math, system and safety engineering there.

psedit; I forgot, I peaked at commaai github repo, the NN code is way above my head and what I could imagine, so I have to give hotz' team that.

Some people really have magic auras.

Actually all my money was on google, i expected them to drive into the scrapfield, after the car companys and Tesla pushed each other into a early market situation and upped Legislation, kicking themselves out of the market, with a carnaggedon. But they backed out, because their managers want the chicken to exist before the egg. So, why not bet on the outsider?

I liked Google's approach a lot. It seemed justly dimensioned, came out of Thrun DARPA success, had years and millions of miles of testing. What's not to love. Also G has the lever to accelerate and change tech if their old approach is obsolete.

Sad they went silent since a year or so.

I said it in other place, including HN, I don't consider Hotz an outsider, but mostly glorified hacker (his attitude with institutions and media don't help me clear this impression); and I don't think hacker mentality bodes well with this task.

Actually, that is a very humanistic point of view. There are third-world countrys where a margin of life-loss is considered "acceptable".

And sorry, but Teslas approach, seems to be a similar hacky "Wing-it with PR-Guns blazing".

About the first point, do countries where such acceptable risk still exist really need self driving vehicles that early ? Could be delayed a bit.

About the second point, see my above comment https://news.ycombinator.com/item?id=12841767

Isn't Google's approach dependent on expensive LIDARs and high resolution 3D maps [0], or have they changed that? I don't see that kind of system scale well.

[0] https://www.youtube.com/watch?v=YXylqtEQ0tk

I've read LIDAR cost was dropping fast, it was expensive at the time and as a short production item; nowadays it's even small.

About the 3D mapping you're probably right, I somehow expected it not to be a deep problem.

> in a conf, hotz mentioned working on time based depth inference instead of the traditional stereo camera setup

This sounds like structure from motion, SfM, which isn't new at all.

https://en.wikipedia.org/wiki/Structure_from_motion

Sure, nothing is new in that field. Space did all that decades ago, except their habits are just not "idiomatic silicon valley".

> I was really hoping it was actually a C64 reference, namely: "where the 'drive' goes"... ,8,1

Actually, the drive number is the “8”; the “1” is a flag which means to load the program to the same address in memory it was saved from, instead of to the default BASIC program start address.

In all fairness...

That's what I was saying. When you use the drive(8), it automatically goes where its supposed to(1).

I believe Arthur C. Clarke said something that's relevant here, "technology that's indistinguishable from magic is probably bullshit."

No, that's not right...

I would include an (unlikely) 3: Hotz really, really does not want to be in the same room as the Feds (he is a confessed hacker) or spend time dealing with them. Maybe more likely: he can sell a ton _now_ in China, had a partner car company, and expect that having those work there would make his success in US either negligible in comparison, or faster thanks the example.

bonus points for mention of c64

LOAD "",8,1 or even more old-school LOAD "",1,1

1,1 !? - Oh hell no. I'm not waiting around for your stupid tape. Get a floppy drive like the cool kids.

This thread makes me happy.

Wow. When I posted this, the only replies were about the Commodore line. Now there's an actual conversation going on!

Reply on news.ycombinator.com