Autopilot Test Fail?

[xav-]

Stationary objects is a major limitation of autopilot.. it doesn’t mean autopilot is useless.. you just have to keep your hands on the wheel and be attentive.. an easy way is to keep good following distance so that you have time to do something (just like you should anyway)

The model s in this video by the way at least attempted to stop which is good. I am pretty sure an AP 10 car would not have applied the brakes at all. Also had the car been upside down or something (not easily recognizable), I suspect autopilot would not have applied the brakes.

The journalist kind of cherry picked their stunt. There was no traffic on each sides and the Mercedes dodged the car with no braking. Had the Mercedes significantly braked this would have been avoided.

If you use autopilot as designed this is not really a problem. The real problem is whether the current hardware can bring FSD. That’s what the problem is.

 

[McRat]

2 seconds of following distance normally stops such collisions. Seriously. It's enough to bring the car to a dead stop, or reduce to the impact to a walking speed. Assuming there is a 2 car lengths between the lead car and the stopped car when you can detect the stopped car, and your normal R/T is .25 seconds (not particularly quick) and .25 to hit the pedal, you have 150+ feet at 60mph. That's conservative assuming the car in the lead is a Kamikaze driver.

The faster you are going, more time you have since R/T is constant, but speed increasing distances and options.

 

[Dr. J]

One of GM's claim to fame is some of the world's best automotive embedded systems engineering.

Better than their work with ignition switches, I guess.

 

[vinnie97]

Considering the amount attention given to these systems and the lack of concrete evidence consumers can use to judge them --- I suspect media bias --- not dissimilar to the media coverage of smartphones --- Samsung is the most underrated smartphone manufacturer of all time --- the only manufacturer to achieve IP68 waterproofing rating, includes medical-grade SP02/hear-rate sensors, retains headphone jack, retains expandable memory storage (can use 256GB microSD), achieves highest body-to-screen ratio without a notch, AND --- can upgrade your phone's OS AND INCREASE THE PERFORMANCE WITH "OUTDATED" 2 year old phones!!! (S7 Edge with the new Oreo OS --- my old phone got FASTER!!)

I'm not buying a Samsung again but LG has done several of those things and have announced Oreo upgrades for the 2-year-old G5 and V20. They went one step further and kept removable batteries through the end of 2016 on the V20 (but sadly succumbed to the industry trend last year). They also included a high-end DAC on the V30 for lossless audio playback.

 

[snakeopus121]

I'm not buying a Samsung again

Yeah Samsung makes the best smartphone screens (and cameras), the only to retain expandable storage, headphone jack, the only to retain performance for older phones with new OS....Yeah I will be avoiding Samsung too /s

Sarcasm aside --- LG sucks at making smartphone screens (or smartphones in general, considering how big of a role screens play in smartphone use)

oh yeah --- IP68 waterproofing --- who is the only manufacturer to achieve this? I forgot...silly me

 

[Knightshade]

oh yeah --- IP68 waterproofing --- who is the only manufacturer to achieve this? I forgot...silly me

...confused why you think only 1 has done this? Sony, LG, and HTC for example all make IP68 rated phones.

Then again unless you do a lot of fresh water scuba diving with your phone IP68 is pretty irrelevant... IP67 is good enough for 3.3 feet of water for 30 minutes- when is your phone going to be in deeper water longer in any sort of normal use?

 

[daktari]

So this was the BBC video, showing us that AEB works but did not stop the car from 38 mph.
What we need pretty soon is a systematic test of the emergency braking system similar to what EuroNCAP does.
The BBC test should also be done with other TACC systems to see performance from the competition.

 

[Knightshade]

So this was the BBC video, showing us that AEB works but did not stop the car from 38 mph.
What we need pretty soon is a systematic test of the emergency braking system similar to what EuroNCAP does.
The BBC test should also be done with other TACC systems to see performance from the competition.

AEB isn't supposed to stop the car from 38mph.

The manual is pretty clear on what the feature does- it intends to only engage when it thinks a collision can not be avoided, and reduces speed by 25 mph to lessen the impact.


AEB and TACC are different things. All model 3s have AEB, only some have TACC. The intention and features of the two systems are not the same.

 

[daktari]

AEB isn't supposed to stop the car from 38mph.

The manual is pretty clear on what the feature does- it intends to only engage when it thinks a collision can not be avoided, and reduces speed by 25 mph to lessen the impact.


AEB and TACC are different things. All model 3s have AEB, only some have TACC. The intention and features of the two systems are not the same.

And that is AEB performance well below the competition.
On the BBC-test, TACC has no role when the lead car changes lane, it is all AEB imo.

 

[Knightshade]

And that is AEB performance well below the competition.
On the BBC-test, TACC has no role when the lead car changes lane, it is all AEB imo.

How do you figure that? Here's a story from only about 6 months ago-

Tesla one of four manufacturers meeting NHTSA automatic braking goals

In its latest report on automatic emergency braking (AEB) standards for automakers, the National Highway Traffic Safety Administration (NHTSA) and the Insurance Institute for Highway Safety (IIHS) have announced that Tesla is one of only four manufacturers so far that has reached the goal of supplying AEB on more than half of its produced vehicles in model year 2017.

or

Guide to Automatic Emergency Braking

Which Brands Do AEB Best?
Most Satisfying: Infiniti, Tesla, and Subaru owners were the happiest with AEB, with more than 74 percent of vehicles rated at very satisfied

 

[daktari]

How do you figure that? Here's a story from only about 6 months ago-

Tesla one of four manufacturers meeting NHTSA automatic braking goals



or

Guide to Automatic Emergency Braking

You are aware that IIHS only tests AEB at 12 mph and 25 mph? While EuroNCAP tests at speeds a lot higher (until the car can no longer avoid impact)? As long as Tesla is not going to be challenged by a stricter test from IIHS I guess they will not bother to improve their AEB. If BBC ran the same test again at 25 mph, I guess the Tesla might even avoid the collision - depending on when the lead car pulled over.

But of course, it seems to me the car was braking - I can not say if the driver braked while the car would just continue onward.

 

[daktari]

You are aware that IIHS only tests AEB at 12 mph and 25 mph? While EuroNCAP tests at speeds a lot higher (until the car can no longer avoid impact)? As long as Tesla is not going to be challenged by a stricter test from IIHS I guess they will not bother to improve their AEB. If BBC ran the same test again at 25 mph, I guess the Tesla might even avoid the collision - depending on when the lead car pulled over.

But of course, it seems to me the car was braking - I can not say if the driver braked while the car would just continue onward.

Forgot: Front crash prevention tests

And check Leaf test EuroNCAP on youtube.

 

[Knightshade]

Forgot: Front crash prevention tests
.

So... your link... it has a link in it to see the online ratings their tests produced.

The model 3 isn't rated yet- but the S and X are.

Ratings

Just put in tesla model s

5 stars, the highest possible score in every test.

So...I'm not sure your link made the point you were hoping to make about how they're "behind their competition"

 

[daktari]

So... your link... it has a link in it to see the online ratings their tests produced.

The model 3 isn't rated yet- but the S and X are.

Ratings

Just put in tesla model s

5 stars, the highest possible score in every test.

So...I'm not sure your link made the point you were hoping to make about how they're "behind their competition"

Yes it did, but one need to have an open mind to see it. Because those 5 stars are based on performance at 12 mph (19 km/h) and 25 mph (40 km/h). So BBC showed it failed to stop at 38 mph (61 km/h). While the new S-class manages to stop from 60 mph (100 km/h) in this commercial video 1 and the Nissan Leaf stops at 40 km/h and 70 km/h with moving front car in EuroNCAP video 2. Volvo XC60 stops from 50 km/h and 70 km/h in video 3.

My point is that Tesla should send their cars for an independent EuroNCAP test of the AEB system before they claim the car to be "the safest".

Video 1

Video 2

Video 3

 

[Knightshade]

Yes it did, but one need to have an open mind to see it. Because those 5 stars are based on performance at 12 mph (19 km/h) and 25 mph (40 km/h). So BBC showed it failed to stop at 38 mph (61 km/h)

It didn't "fail" to stop... it's not supposed to stop.

Teslas AEB is designed to reduce speed by 25mph. After that either the crash is already over, you've avoided the crash, or you have had time to manually intervene in whatever method is needed.... (or AEB somehow triggered on a false positive- in which case you don't WANT it to keep braking anyway).

That's why it's an EMERGENCY system.

If you want a mode where the car will routinely and complete stop itself from 60 mph if a car in front slows down to a stop you want to be using TACC- not relying on a last-ditch emergency system.[/QUOTE]

 

[daktari]

It didn't "fail" to stop... it's not supposed to stop.

Teslas AEB is designed to reduce speed by 25mph. After that either the crash is already over, you've avoided the crash, or you have had time to manually intervene in whatever method is needed.... (or AEB somehow triggered on a false positive- in which case you don't WANT it to keep braking anyway).

That's why it's an EMERGENCY system.

If you want a mode where the car will routinely and complete stop itself from 60 mph if a car in front slows down to a stop you want to be using TACC- not relying on a last-ditch emergency system.

Ah, and I thought you also would like a system that avoided a crash instead of just reducing the speed before the crash. Well, we are all different!

 

[Dr. J]

Well, we are all different!

So saith the wise one:

 

[Knightshade]

Ah, and I thought you also would like a system that avoided a crash instead of just reducing the speed before the crash. Well, we are all different!

I would, that's why I'm getting EAP with mine. Which, unlike AEB, is actually designed to maintain a specified safe distance from cars in front of it at all times, up to and including bringing the car to a complete stop when needed.

Unlike AEB, whose actual stated purpose is to reduce the damage in an EMERGENCY situation where a crash appears to have become unavoidable.

Right tool for the right job and all that.


I'm actually having a good bit of trouble imaging a real world situation where needing it to stop to 0, above 25, versus the current behavior would be genuinely useful.

Generally if you're going over 25, and aren't using TACC, and are paying so little attention you don't notice a stopped car up ahead, and don't notice any of this including the car you're in rapidly decelerating on its own from AEB by 25 mph to prompt you to hit the brakes yourself, you're probably already dead at the wheel anyway right?

I mean... I guess that feature would be useful if you'd passed out still alive while driving and didn't have TACC on. So, ok, there's your one use case I suppose.

Seems fairly narrow.

Doubly so when I can't imagine many situations I'd ever be going over 25 and not using TACC. Which, again, stops all the way down to 0.