Scenarios in which Tesla FSD Beta 9.0 fails • Hacker News Board

Re: Scenarios in which Tesla FSD Beta 9.0 fails

#411

Post by matz1 » Tue, Jul 13, 2021, 2:45 AM UTC

cogman10 wrote: post #398

matz1 wrote: post #396

Earlier quoted context omitted.

Well see, I personally accept and support far from perfect no driver at wheel car. I doubt I'm a minority.

You say that now, will you say that when (not if) a driverless car kills a kid or animal that a human wouldn't? Were you ok with Uber's killing?

Human kills already, and yet certainly do not want human to be banned for driving.

I too expect and accept that some accident caused by driverless car that kills kid/animal (that human wouldn't do) will happen (if not already happen). To expect otherwise will be unrealistic.

Re: Scenarios in which Tesla FSD Beta 9.0 fails

#412

Post by amanaplanacanal » Tue, Jul 13, 2021, 2:47 AM UTC

marvin wrote: post #339

Invictus0 wrote: post #312

Earlier quoted context omitted.

> one accident for every 4.19 million miles driven Not really a great metric. We should be looking at disengagements per MM.

If you're evaluating the system as a L5 autonomous driving system. Not if you're evaluating it as a safety assistance feature.

If it’s a safety assistance feature, it’s not full self driving is it? My crosstrek comes with plenty of safety assistance features too.

Re: Scenarios in which Tesla FSD Beta 9.0 fails

#413

Post by QuibbleQuota » Tue, Jul 13, 2021, 2:50 AM UTC

jazzyjackson wrote: post #232

Theodores wrote: post #184

Earlier quoted context omitted.

A century ago horses had Full Self Driving with the human having to be vigilant and in charge. The horse could go crazy in scary situations endangering others but you could train that out and make a good war horse. Basic stuff like following the road around a bend was something the horse could do but you could not tell it to autonomously go across town to pick someone up. People were okay with horses and knew what to…

Horses could at least make their own way back home, and they were self-fueling to boot!

Definitely not zero emissions though…

Re: Scenarios in which Tesla FSD Beta 9.0 fails

#414

Post by amanaplanacanal » Tue, Jul 13, 2021, 2:51 AM UTC

the8472 wrote: post #238

foepys wrote: post #165

Earlier quoted context omitted.

> People will die today because a driver was drunk/distracted/suicidal/rode raged or had a medical problem. That's a straw man argument. Mandatory breathalyzer tests before every engine start and yearly medical and psychological check-ups would solve a lot of these problems as well. > It gets safer overtime. There is no evidence of this happening yet. It's still trying to drive people into concrete pillars. Remember…

> It's still trying to drive people into concrete pillars. That in itself doesn't constitute evidence that it isn't getting safer. After all humans occasionally drive into concrete pillars too. What matters is that the distance traveled without incident goes up in a super-linear fashion over time. It would be nice if someone had hard numbers on that.

I’d say if we compare Tesla self driving numbers with human drivers driving similar cars, and count every disengagement as an accident, it would be a fair comparison. Tesla wouldn’t come out looking very good with those numbers though.

Re: Scenarios in which Tesla FSD Beta 9.0 fails

#415

Post by GhostVII » Tue, Jul 13, 2021, 2:54 AM UTC

cs702 wrote: post #120

Many of the comments here seem a bit... unfair to me, considering that these clips were handpicked. I watched (and fast-forwarded) through a few of the original, full-length videos from which these clips were taken. The full-length videos show long drives (in some cases, hours long) almost entirely without human intervention, under conditions explicitly meant to be difficult for self-driving vehicles. One thing I rea…

For level 5 (which Elon promised), FSD has to go tens of thousands of miles without a collision-avoiding intervention. Right now it can't even do 10 miles.

Cherry picking is completely fine because with the limited number of beta users, even a few incidents is enough to show that FSD isn't nearly ready for level 5.

And it's pretty easy to make a car that can self drive on simple roads without a tonne of traffic, so being able to do that without intervention isn't much of an accomplishment.

Re: Scenarios in which Tesla FSD Beta 9.0 fails

#416

Post by pas » Tue, Jul 13, 2021, 2:55 AM UTC

bumby wrote: post #279

flowerlad wrote: post #222

Earlier quoted context omitted.

Why not have redundant sensors and crosscheck them? Boeing 737 Max had only one AOA sensor [1], and that wasn't a great idea. https://www.cnn.com/2019/04/30/politics/boeing-sensor-737-ma...

> Boeing 737 Max had only one AOA sensor Just a small nit-pick but it makes the case against Boeing worse. The airframe had multiple AOA sensors but the base software only used one sensor reading. Note the image in [1] shows readings from both a "left" and "right" AOA. From your link: > software design for relying on data from a single AOA sensor Boeing sold a software upgrade to read both AOA devices. (This still le…

The high level problem was the whole mess that Boeing and the FAA forced themselves into by lying to pilots about the plane being like the old plane. This motivated Boeing to solve everything in a sort of clandestine way, which meant there was no proper error handling, and Boeing simply covered their asses by pointing to the runaway stabilizer checklist, which they claimed was sufficient to handle MCAS problems.

This should never have been greenlighted, approved, manufactured, designed. This went against a culture of safety (least surprise, etc) at every level.

Re: Scenarios in which Tesla FSD Beta 9.0 fails

#417

Post by foobiekr » Tue, Jul 13, 2021, 2:58 AM UTC

Theodores wrote: post #184

Syonyk wrote: post #17

Earlier quoted context omitted.

The history of human machine interactions makes it clear that if you have the human as the primary control (in the main decision loop, operating things) and the automation watching over their shoulders, it works pretty well. The opposite - where the automation handles the common case, with the human watching, simply does not work. It's not worked in aviation, it's not worked in industrial systems, it's not worked in…

A century ago horses had Full Self Driving with the human having to be vigilant and in charge. The horse could go crazy in scary situations endangering others but you could train that out and make a good war horse. Basic stuff like following the road around a bend was something the horse could do but you could not tell it to autonomously go across town to pick someone up. People were okay with horses and knew what to…

Horses do not run into concrete pillars.

Re: Scenarios in which Tesla FSD Beta 9.0 fails

#418

Post by GhostVII » Tue, Jul 13, 2021, 2:59 AM UTC

I don't know why Tesla is so focused on self driving in cities. Seems like a far better use of time would be to get to level 3 self driving on highways (which is what other automakers seem to be trying to do). That gets you most of the benefits of self driving (not having to pay attention for long trips), with probably 1% of the complexity of city driving.

Getting city driving working just seems so impossibly hard, there is no way you are going to be able to use ML to determine road rules since they vary by jurisdiction and over time, and there are so many edge cases you won't have enough data for. But how do you write a program that can read a street sign, interpret it's meaning, and act upon it safely?

Even if Tesla just got autopilot to the point where it could be level 3, where drivers don't have to pay attention, that would be a massive improvement, and so much easier than FSD.

Maybe they painted themselves into a corner though with all Elon's blatant lies about FSD, can't stop development without a lawsuit. They seem kind of screwed anyways, looks like they are running into the limits of the current camera positioning and onboard processing power, might need to do another iteration of replacing every FSD owners hardware.

Re: Scenarios in which Tesla FSD Beta 9.0 fails

#419

Post by nuker » Tue, Jul 13, 2021, 3:04 AM UTC

TheParkShark wrote: post #348

raisedbyninjas wrote: post #314

Earlier quoted context omitted.

They currently charge $10k for FSD. Even if they don't drop any of their cameras and processing equipment, $500 LIDAR is a 5% increase. Is it just me or are we politely ignoring Musk's reluctance to admit his bold proclamations against LIDAR are the biggest factor in their design decisions.

Andrej Karpathy already explained they were getting too much noise from Radar. Adding LIDAR would only complicated it. Unless you have better knowledge than Andrej Karpathy, I’ll go with his word on this. Edit: nitrogen I can’t comment directly on yours, so I’ll ass it here. That’s a good goal to have and they tried that for well over two years. They found too much noise and confusion. If Radar says one thing and Vis…

> Adding LIDAR would only complicated it.

More sensors is better. They just have not yet learned how to merge data. Example is drone autopilot, that have IMU and GPS and uses Kalman filtering.

Re: Scenarios in which Tesla FSD Beta 9.0 fails

#420

Post by aeternum » Tue, Jul 13, 2021, 3:09 AM UTC

Syonyk wrote: post #241

moralestapia wrote: post #181

Earlier quoted context omitted.

Here's what happened another time someone argued that "radar can't detect stationary objects". https://news.ycombinator.com/item?id=16722461 Radar does detect stationary objects, come on.

Yes, radar can easily detect stationary objects. Tesla's implementation of their radar processing drops all returns from anything stationary, given the forward speed of the car. In the context of a Tesla thread, I didn't think that needed to be specifically stated, but will do so in the future.

A big problem with radar was the resolution. The radar signature of a highway overpass looks extremely similar to the monorail column.

Emergency breaking on a highway when going under an overpass is also a huge problem, so radar may not be the answer.