Tesla Live Stream – Autonomy Day [video] • Hacker News Board

Re: Tesla Live Stream – Autonomy Day [video]

#511

Post by Upvoter33 » Tue, Apr 23, 2019, 5:43 PM UTC

p1esk wrote: post #360

sytelus wrote: post #336

Earlier quoted context omitted.

Karpathy is great at presenting executive-101. However self-driving is much much more than image classification. You have a time component, you need to integrate information as it comes in to meaningful state (aka SLAM) and make sequential decisions that has potential to kill people. This is still unsolved research problem unlike punny ImageNet classification. Karpathy could have easily spent entire day talking about…

This is still unsolved research problem Luckily, Karpathy is pretty good at solving research problems. Karpathy could have easily spent entire day talking about current state of the art and how they have improved on it Why would he do that? To help Uber? Collecting data is actually more trivial part of this problem. For Tesla, yes. For all others that’s the hardest part.

> Luckily, Karpathy is pretty good at solving research problems.

Oh come on - there are a lot of people good at solving research problems, but hard problems remain unsolved. To pretend driving autonomously in the real world is anything but very very hard is naive.

Re: Tesla Live Stream – Autonomy Day [video]

#512

Post by Fricken » Tue, Apr 23, 2019, 6:41 PM UTC

leesec wrote: post #358

Fricken wrote: post #173

Earlier quoted context omitted.

Vision is probablistic. You can improve your probabilities with more data and deeper nets, but the probability is never 100%. Inferring depth using triangulation of extracted features using vision is also probabilistic. Lidar gives you very rudimentary depth and positional data, but it is precise, and reliable, and the false positives are low. Lidar is not seeing the world as a series of probabilistic phantasms that…

>"We've gone over this multiple times. Are we sure we have the right sensor suite? Are we sure we have everything we need?.... No." Where did he say this? He repeatedly said they have everything they need to drive. >but before you can deploy, you need to integrate Lidar, Why? >not all the conventional things you need to build a complete autonomous vehicle. No one knows how to make a completely autonomous vehicle beca…

There's a reason why nobody else in the industry thinks they can make robotaxis work without Lidar. Of course, if you were familiar with the rest of the industry you wouldn't say Tesla has gotten farther than anybody else in the field, that's just ignorant.

Re: Tesla Live Stream – Autonomy Day [video]

#513

Post by imtringued » Tue, Apr 23, 2019, 6:48 PM UTC

selectodude wrote: post #112

DeonPenny wrote: post #64

Earlier quoted context omitted.

I keep saying this he knows what he doing. You can tell he knows what's happening. Which is rare for a CEO to know this level of detail.

Which might explain why he's a crummy CEO. If your job is to strategize for the entire company and you're busy learning the minutiae that you spend a lot of money for other people to worry about, maybe you need to consider using your time a little better. If you want to build neural nets, keep your stock, fire yourself and go build neural nets.

Actually being aware of the technical details of your core business is a very useful skill to have as a CEO. But it doesn't seem to actually result in any benefits with Tesla. He probably knows the limitations of neural networks but that doesn't stop him from using the tried and true startup strategy of promising the world and delivering very late. Building a car that can stop on red traffic lights is something they should have had by 2016 already. But it's 2019 and their self driving technology they have demonstrated in their demo blatantly ignores the speed limits on the highway by roughly 10mph.

Re: Tesla Live Stream – Autonomy Day [video]

#514

Post by DeonPenny » Tue, Apr 23, 2019, 7:06 PM UTC

dforrestwilson wrote: post #421

DeonPenny wrote: post #373

Earlier quoted context omitted.

Yes absolutely saying Tesla who gets camera data from it's a half million car doesn't give it an advantage is crazy. That not even including the fact it's the only company who can do its strategy. Google would need to get constant data and GM and legacy automakers would need sensor suites on all it's cars yesterday. No one knows if Tesla strategy will work because they don't have the data collection in place.

Neither does Tesla which makes it a moot point. They have no way to store or transmit the massive data you are describing off the platform do they? My understanding is that they have very limited storage and transmit onboard.

Based on their talk today and Andrew previous talk where he shows explicitly tools that do just that download data constantly is exactly what they do. https://vimeo.com/274274744

I mean saying a phone can upload videos to youtube but a can can't to tesla is a weird ledge to stand on. Even their windshield wipers work based on sending video data to tesla to be learned on.

Re: Tesla Live Stream – Autonomy Day [video]

#515

Post by 1123581321 » Tue, Apr 23, 2019, 7:17 PM UTC

VikingCoder wrote: post #496

1123581321 wrote: post #427

Earlier quoted context omitted.

The person who wrote that doesn’t realize how fast 140 would seem. That kind of speed differential is not something you can attempt to avoid. Even 100 vs. 70 is rare, and the safest thing to do given that differential is nothing. This is not an area any car company should worry about.

I agree that the safest thing to do is nothing. Which is why if I'm about to change lanes, and I spot this kind of behavior coming up rapidly from behind me, I abort my lane change. Even if I was moving to the Right lane (you never know which way these jerks are going to go.) It is absolutely something a car company should worry about. Avoiding obstacles from any direction. If a power line is falling down. If a const…

I see. I would also consider an aborted lane change doing nothing. However, I wouldn’t try to jerk back into my lane if a car was approaching me with that kind of differential since a) they don’t have time to react and may have been planning to go into my origin lane, for all I know, and b) by the time such a fast car is spotted, they would likely pass my car too quickly for any lateral movement to leave their trajectory.

Your other examples would be happening in front of the car.

Re: Tesla Live Stream – Autonomy Day [video]

#516

Post by greedo » Tue, Apr 23, 2019, 7:48 PM UTC

s_y_n_t_a_x wrote: post #432

greedo wrote: post #251

Earlier quoted context omitted.

The basics. Like having a trunk lid design that sheds water properly? https://teslamotorsclub.com/tmc/threads/water-dripping-into-...

Shut down the company the trunk is slightly off. Tesla is new to the game, give them a few more cars to catch small stuff like that.

Tesla has been around for 16 years, so I think it's not too critical to expect them to be able to sweat the small stuff.

Re: Tesla Live Stream – Autonomy Day [video]

#517

Post by lonnyk » Tue, Apr 23, 2019, 7:56 PM UTC

mdorazio wrote: post #500

lonnyk wrote: post #477

Earlier quoted context omitted.

The tech specs of AGX Pegasus say they're 130 TOPS[1]. Or am I misunderstanding something? [1] https://developer.nvidia.com/drive/drive-agx

Where are you getting 130 from that page? I don't see it. On this other page [1], it clearly says 320. "NVIDIA DRIVE AGX Pegasus™ achieves an unprecedented 320 TOPS" [1] https://www.nvidia.com/en-us/self-driving-cars/drive-platfor...

If you click “tech specs (click to expand)” then it says it on there.

Thanks for your link though. I couldn’t find 320 anywhere

Re: Tesla Live Stream – Autonomy Day [video]

#518

Post by cromwellian » Tue, Apr 23, 2019, 8:00 PM UTC

salty_biscuits wrote: post #478

cromwellian wrote: post #474

Earlier quoted context omitted.

Neural networks are probabilistic. They output predictions and are rarely 100% confident. So when Tesla's visual neutral networks are assigning inferred labels to things like cars, bikes, pedestrians, road barriers, etc, there is always some chance, say, 0.1% that it's wrong. If you have other sensors, like LIDAR, radar, and high resolution maps, then you can corroborate your predictions, so for example, if the visua…

Ok, lots of separate issues there. In terms of actually using a map I think my original point still stands. You need to use online navigation information to localize yourself within the map before you can use it. There is also the non trivial issue of actually doing a bayesian update of "can I go around this corner" given the error budget coming from how accurate is the map, how accurate is my position within the map…

Visual systems have problems with obstructions and adversarial imagery, having side facing radar which can go under and around obstacles at long distances for example is a huge benefit at intersections.

Musk swept this simple fact under the rug when asked during the press conference, again overselling a 100% visual solution.

Re: Tesla Live Stream – Autonomy Day [video]

#519

Post by gamblor956 » Tue, Apr 23, 2019, 8:36 PM UTC

justapassenger wrote: post #325

grecy wrote: post #287

Earlier quoted context omitted.

You could, but you're not a billionaire running multiple companies that have already done a ton of things that many, many experts said could not be done. Also note, I'm not saying Elon will do all or even half of them. I'm saying it's going to be damn exciting to watch, and I'm keeping my fingers crossed that he does, because I'd like the world to get better, and I'm happy to be an optimist.

> You could, but you're not a billionaire running multiple companies that have already done a ton of things that many, many experts said could not be done. Like what? I often hear people saying that experts were saying it's impossible to land a rocket or make a good EV car. But I didn't hear experts saying that. What I did hear, is experts saying it's impossible to make it in cost effective way. And for that - econom…

I think SpaceX will stick around as long as Shotwell remains COO. She's done a great job of running it. All of SpaceX's disasters can be traced to the pre-Shotwell period, or to Elon interfering in her management (such as this past weekend's rushed crew capsule test).

Re: Tesla Live Stream – Autonomy Day [video]

#520

Post by VikingCoder » Tue, Apr 23, 2019, 9:02 PM UTC

1123581321 wrote: post #515

VikingCoder wrote: post #496

Earlier quoted context omitted.

I agree that the safest thing to do is nothing. Which is why if I'm about to change lanes, and I spot this kind of behavior coming up rapidly from behind me, I abort my lane change. Even if I was moving to the Right lane (you never know which way these jerks are going to go.) It is absolutely something a car company should worry about. Avoiding obstacles from any direction. If a power line is falling down. If a const…

I see. I would also consider an aborted lane change doing nothing. However, I wouldn’t try to jerk back into my lane if a car was approaching me with that kind of differential since a) they don’t have time to react and may have been planning to go into my origin lane, for all I know, and b) by the time such a fast car is spotted, they would likely pass my car too quickly for any lateral movement to leave their trajec…

> I would also consider an aborted lane change doing nothing.

Aborting a lane change before you initiate it is changing your behavior, in response to a danger.

> Your other examples would be happening in front of the car.

Or from the side.

Or from behind, again. If I'm in stop-and-go, danger can come from behind. Picture an ambulance coming up from behind. You absolutely should change your behavior, depending on conditions all around you. Not just the front.