Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Tesla footage of braking before crash ahead [video]

youtube.com

Post reply on HN
Print view View original

351–360 of 386 posts

Re: Tesla footage of braking before crash ahead [video]

#351

Post by clebio »

We don't need to speculate as to the speed or proximity of the two vehicles ahead. The data the autopilot system used has it. Does Tesla own that data, or the vehicle-owner? It's collected on the vehicle, which is _owned_. I'm sure Tesla gets a copy -- do they retain the rights as well?

I would not be surprised if Tesla were subpoenaed (or otherwise requested of) for the data, in a case like this (similar to the recent Amazon Echo news, or cellular providers with geolocation in abduction cases).

Re: Tesla footage of braking before crash ahead [video]

#352

Post by greglindahl »

tajen wrote: post #349
notheguyouthink wrote: post #264

Earlier quoted context omitted.

> It's the sort of thing you could, however, program into a robot and expect the robot to adhere to it. I've often wondered what people will initially do when being driven around by a robot. How anxious will it make them feel, not to give up control, but to see a robot driving "so slowly". Not going 10 over, letting other drivers in when possible, always slowing to keep a safe distance, not passing aggressively if so…

The latest OS upgrade of the Tesla prevents the autopilot from going faster than the speed limit. I wonder what owners think about that.

I think that you were fooled by the news coverage. It's only on non-divided highways that the new restriction applies, and there was already a restriction to speed limit + 5mph on such roads. On divided highways, you can set any speed up to 90mph.

Here's an accurate article, which (alas) still has an inaccurate headline: https://electrek.co/2016/12/22/tesla-autopilot-speed-limits/

Re: Tesla footage of braking before crash ahead [video]

#353

Post by greglindahl »

dharma1 wrote: post #325
ge96 wrote: post #60

do the cars have some serious computer hardware or is it really good code/cloud assisted computation? Maybe ANN's active while driving?

New Teslas have Nvidia PX2 which is very fast. Can easily run deep neural networks realtime https://blogs.nvidia.com/blog/2016/10/20/tesla-motors-self-d... Not sure what HW their previous cars had

The previous Tesla had a Tegra 4, which also powered the 17" touchscreen, and a Tegra 3 powering the instrument panel.

Re: Tesla footage of braking before crash ahead [video]

#354

Post by HappyTypist »

clebio wrote: post #351

We don't need to speculate as to the speed or proximity of the two vehicles ahead. The data the autopilot system used has it. Does Tesla own that data, or the vehicle-owner? It's collected on the vehicle, which is _owned_. I'm sure Tesla gets a copy -- do they retain the rights as well? I would not be surprised if Tesla were subpoenaed (or otherwise requested of) for the data, in a case like this (similar to the rece…

If you're in the EU, you can get this data.

Re: Tesla footage of braking before crash ahead [video]

#355

Post by pacificmint »

loeg wrote: post #283
Tloewald wrote: post #230

Earlier quoted context omitted.

The SUV was pushed into the crash barrier and small car wasn't, so it's a highly unfair comparison. We did a lot of safety research before buying our current round of cars, and you need to not generalize about cars based on classifications like "SUV". Look at actual data from actual cars on actual roads, e.g. http://www.iihs.org/iihs/topics/driver-death-rates Note that you can find vehicles like the Honda Pilot that…

> What's horribly unsafe to everyone is passenger cars built on truck frames. Have any examples? The vast majority of new SUVs (which the one in this incident appears to be) are unibody rather than body-on-frame ("truck frame" style).

> Have any examples? The vast majority of new SUVs [...] are unibody rather than body-on-frame

Apparently there are 18 models of Body-On-Frame SUVs that are being sold in the US in 2016:

  Cadillac Escalade
  Cadillac Escalade ESV
  Chevrolet Suburban
  Chevrolet Tahoe
  Ford Expedition
  GMC Yukon
  GMC Yukon XL
  Infiniti QX80
  Jeep Wrangler
  Land Rover LR4 *
  Lexus GX460
  Lexus LX570
  Lincoln Navigator
  Mercedes-Benz G-Class
  Nissan Armada
  Toyota 4Runner
  Toyota Land Cruiser
  Toyota Sequoia
Source: http://www.goodcarbadcar.net/2016/02/body-on-frame-suv-produ...

Re: Tesla footage of braking before crash ahead [video]

#356

Post by GavinMcG »

mikeash wrote: post #336
GavinMcG wrote: post #332

Earlier quoted context omitted.

Go back and take a gander at the question mark in the comment you're criticizing, and the lack of one in the comment I'm addressing.

Are you thinking I said your comment is a strident claim that requires confrontation to disagree with? I said the opposite: both your comment and the one you originally replied to are conversational and easily amenable to contrary views.

No, the other way I could disagree. I thought the first one was the closed-off claim.

Re: Tesla footage of braking before crash ahead [video]

#357

Post by brianwawok »

pacificmint wrote: post #355
loeg wrote: post #283

Earlier quoted context omitted.

> What's horribly unsafe to everyone is passenger cars built on truck frames. Have any examples? The vast majority of new SUVs (which the one in this incident appears to be) are unibody rather than body-on-frame ("truck frame" style).

> Have any examples? The vast majority of new SUVs [...] are unibody rather than body-on-frame Apparently there are 18 models of Body-On-Frame SUVs that are being sold in the US in 2016: Cadillac Escalade Cadillac Escalade ESV Chevrolet Suburban Chevrolet Tahoe Ford Expedition GMC Yukon GMC Yukon XL Infiniti QX80 Jeep Wrangler Land Rover LR4 * Lexus GX460 Lexus LX570 Lincoln Navigator Mercedes-Benz G-Class Nissan Arm…

Those are mostly the biggest ones. Maybe 25% of market at best?

Re: Tesla footage of braking before crash ahead [video]

#358

Post by sytelus »

gingerlime wrote: post #105
sytelus wrote: post #67

This is great video to study on many levels beyond self-driving car. Some observations: * The car behind didn't show turn signal so probably driver wasn't paying attention. * A small car escaping with not too much damage while SUV rolls over many times * From physics point of view, the car behind produced almost perfect torque that started horizontal spin on SUV * Collisions with barrier at high speed can produce lot…

I'm feeling rather dumb, but can't really intuitively understand the physics here. I see a small car hitting a pretty huge one and yet spins the huge one several times. I always thought those bigger cars are safer / stronger / more resistant to hits? Is it purely because it hit it at the corner, so all the force went into the "spinning axis"? (did I say I feel dumb?)

Torque is

r X F

meaning that small force applied over a long length can produce huge torque! You can move a huge rock if you use a long enough metal stick for leverage. This is why Archimedes said,

Give me a lever long enough and a fulcrum on which to place it, and I shall move the world.

Large angular acceleration due to large torque is then directly result of Euler's equations.

Re: Tesla footage of braking before crash ahead [video]

#359

Post by bdhess »

gohrt wrote: post #312
bdhess wrote: post #289

Earlier quoted context omitted.

The 26% statistic is misleading in this case-- just because the death didn't involve truck driver "error" doesn't mean a different movement pattern for the truck wouldn't have made a difference. For example, in the original video, if the Tesla autopilot hadn't braked, and the car had been involved in the crash, it's not clear to me that error on the part of the Tesla driver would have been listed as a cause.

It should be, though. Car #1 stopped shorted and car #2 hit it. Car #2 was at fault for hitting Car #1. Car #3 (Tesla), if it didn't stop, would be at fault for not maintaining stopping distance, hitting Car #2.

You're implying that minimum following distance If Driver #3 can't see Car #1 past Car #2, and Car #2 is traveling at highway speed, I think it's a stretch to say that Driver #3 could be faulted.

[0] http://www.driveandstayalive.com/stopping-distances/

Re: Tesla footage of braking before crash ahead [video]

#360

Post by sytelus »

79d697i6fdif wrote: post #170
sytelus wrote: post #67

This is great video to study on many levels beyond self-driving car. Some observations: * The car behind didn't show turn signal so probably driver wasn't paying attention. * A small car escaping with not too much damage while SUV rolls over many times * From physics point of view, the car behind produced almost perfect torque that started horizontal spin on SUV * Collisions with barrier at high speed can produce lot…

The biggest thing I noticed was just how easily SUV's flip over. This has been a problem forever I'm just surprised that regulators don't take more notice of it. Rollovers are far more likely to result in injury or death. Most cars just don't flip, all the traction their tires have is simply not enough to do it on flat ground. Yeah anything could flip over once you go off the road but the SUV in this looks like a rol…

I was wondering how did SUV rolled because for that to happen you need force getting applied on car that is on Y axis (the +X axis being the forward direction) which will ultimately produce the torque causing the roll. After looking at video at low speed, here's how we got that force vector:

1. Hatchback hits on tangent producing torque and spinning SUV by 90-degree.

2. SUV collides with barrier which produces force vector that is normal to collision surfaces and slightly off forward direction.

3. The driver now seems to have pressed on breaks really hard which produces another force vector. But this vector is now quite off original forward direction because car had been rotating.

So the moral of the story is that if SUV is rotating and you press on breaks really hard, you will get roll. This is purely because SUV has higher clearance and so even small force can produce enough torque that will cause roll. It looks to me if driver hadn't pressed on breaks really hard, the rolls could possibly been avoided or reduced.

Post reply on HN