Separate names with a comma.
Discussion in 'Model S' started by malcolm, Jul 7, 2018.
Tesla Model S fails auto braking test, Tesla questions validity of the test
Agree with the article, why on earth would they:
a) do such a test in front of the press?
b) test a 2015 Model S against a brand new Volvo?
Given the high safety ratings Tesla has achieved, and that Volvo wants to be the brand of safety, I smell all kinds of agenda in this.
Tesla S safety rating is much worse than let's say Mercedes E-Class. (EuroNCAP).
The Tesla appears not to brake at all:
It seems to be some sort of marketing for Volvo "A 2018 Volvo, with Volvo's latest technology is better at emergency braking than a three year old Tesla".
The war continues... an unimportant organization in unimportant country stages an event where it shows how badly a 3 years old car behaves in a situation it was never designed to cope with.
It is done outside of USA so it is harder for Tesla to put up a legal fight.
Expect to see even more of such *sugar*.
Tesla didn’t implement automatic emergency braking until 2017. That’s one of the reasons we bought new (last year) instead of CPO.
The fact this was done in front of a press pool shows it was obviously staged.
That's for AP2. AEB in Ap1 has existed for quite a while.
Tesla Model S AEB is not designed to prevent collisions. So this is normal behaviour.
Model S Owner's Manual page 92:
That’s the way it is designed doesn’t mean it is the ideal outcome. All else equal, a car that can prevent a collusion is better than one that can only lessen the impact.
Of course, it must be done in a way that keeps false positive to a minimum, or it would be so annoying that it would be disabled, or worse trigger accidents on its own. It is possible that the Volvo is much more conservative on triggering Autobrake and that causes problem on its own, w just don’t know.
One cannot test collision avoidance in isolation (like the circus we see here), ability to reject false positive must be considered at the same time.
I had a new Volvo with all the toys on loan recently, had to turn most of it off as the false positives were going to get us killed a lot faster than the potential driving aids were worth...
The fact that only the Volvo is fitted with all the recording gear is a bit telling of this test’s goal...
Then it was designed stupidly.
Not necessarily. AEB is a trade off between false positive and sensitivity. We cannot determine the system’s capability unless false positives are also measured.
Tesla is design to brake only when collision is inevitable, so by definition it can only mitigate impact and not prevent it. This design would greatly reduce the chance that it will not kick in accidentally, so you will keep it on and active. For example, if you are approaching a stopped car at speeds and have the intention to steer around it, it will not AEB. Other systems that are tuned to avoid collision will either need to read your mind, or trigger a lot more false positive braking, slamming on the brakes even when there is another path to avoid the collision.
Tesla’s system puts the driver under control, assisting when it becomes certain that the driver will not be able to avoid the accident.
To me this illustrates a major issue with AEB.
That issue is not all AEB systems are created equal, and we have no idea how the various systems rank against each other. Heck we don't even know what the test setup was, and whether they were following an agreed upon testing methodology.
To someone that didn't know anything about AEB systems they'd assume the Tesla system was bad or didn't work without realizing that a 2015 Model uses the MobileEye system. The MobileEye system is used on a quite a few vehicles. Sure there might be something different about how it was implemented for the Tesla that might have impacted the test.
The other issue is we also have no idea how the false positives compare.
I know the AEB of an AP2 car is likely superior to that of an AP1 car in terms of false negatives. That the AP2 will have fewer false negatives especially with stopped object/vehicles. But, from the user complaints AP2 seems to have higher false positives.
The industry needs testing that is a lot more extensive, and time relevant. How in the world is this useful to me as an owner of a 2015? It might have been useful back in 2015 when I bought the vehicle.
Back in 2015 I knew it kinda sucked, and I had issues with how Tesla worded things. I preferred how Subaru worded how there system worked because it felt a lot more satisfying. Where Subaru claimed the system would always stop if there was less than a 30mph differential (or something like that).
Where Tesla was like "we'll remove 20mph from the speed" if the accident is unavoidable. Then it went on to saying all the different scenarios it might not work.
In actually using it I don't have much issue with it. I like that it stays out of the way, and so far I've never had any AEB false braking. For me the FCW is a lot more useful.
Exactly... AEB will reduce the vehicle speed by 25 MPH and then release the braking control to the driver.
Strange test. I still think Tesla should send a AP2 vehicle to EuroNCAP for these tests. You will find all their tests at YouTube, including AEB for pedestrians, bicycles and rear ending.
What the manual above should, but doesn't say is wether a Tesla ever will brake for either of those above, or a crossing or opposing car. Or only rear ending?