-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Note that MobilEye has played a key role in defining EURO NCAP and IIHS test procedures for AEB systems, and sells to manufacturers primarily by touting the increase in IIHS and NCAP ratings by equipping with a camera-based AEB system.
Remember not long ago, Mercedes had an excellently tested AEB system per IIHS and EURO NCAP and they attempted to stage a real-world demo and ran into "interference" when it wasn't an open testing space with a silver blow-up doll:
EyeSight did do quite well and it's commendable, but again, collision system specs on paper (in terms of what speeds they work and how much they reduce your speeds) are largely unregulated or simplistically tested. Real world performance is hard to quantify, and in my experience AP1 was absolutely wonderful compared to other cars, and AP2 still is quite good.
stopcrazypp
Well-Known Member
I don't believe Eyesight is mobileye. Subaru uses their own stereo camera system (without radar).
Sorry for the reply a while later, but I didn't see anyone else address this: the physics/math doesn't work that way. AEB slows by up to 25 mph, but it can't do that instantly, as that'd require infinite force and an infinite friction coefficient with the ground. Rather, it applies some acceleration(hitting the brakes) over a period of time until the 25 mph cutoff is met.
So, yes, if it actually manages to slow that 25mph, then there'll be no collision. But if it doesn't have enough time to actually accomplish that before hitting the car/whatever in front of it, then the speed of the collision will just be reduced.
The part that's missing is that AEB only begins to brake when it has determined that even with full braking power, an accident is unavoidable. So unless something else changes, such as the car in front speeding up or the driver taking steering action, or your brakes / tires being dramatically better than the AEB model, the collision is not avoidable at the time that braking starts.
Right. I was just pointing out that assuming that you're totally safe if you're going < 25mph is a bad assumption.
It's a mistake to assume that:
1. Tesla isn't working on a reliable AEB
and
2. Work done on these two are mutually exclusive.
It's highly likely that much of the same underlying software is feeding both and that improvements to one lead to improvements in the other.
JohnnyG
Weee!
AEB is for human driven cars, not FSD cars.
FSD is always AB (Automatic Braking).
AEB (on Teslas) is purposely designed to not engage until the last moment, leaving the driving decisions to the driver.
AEB was designed to save humans when distracted. FSD, by the pure nature of it, can not be distracted.
FSD will always be automatic braking; drastically reducing the emergencies in which AEB might engage.
Thus, as the two may cross paths, FSD does not require AEB, because FSD is always in control of the braking; engaging way before an eminent crash is detected in most situations where it would be activated.
As @JohnnyG wrote, AEB is not for FSD vehicles. The algorithm is completely different and actually AEB is, in many ways, more difficult. AEB has to step in well after FSD would have braked. It has to also have a slew of overrides programmed into it to allow a human driver to make the final driving decisions. Much of the work with AEB has nothing to do with FSD.
so i emailed tesla again and they claim the info is correct for HW2. according the CR they "just got this info"
That statement has a lot of assumptions. What do you mean by "can't"? Again, AEB is more work beyond FSD and different from FSD. It is a completely different code path. If they have FSD working, they would still have to spend R&D time and resources to get AEB to work, leaving the FSD code path aside.
I don't need to test it........ Even if it work FLAWLESSLY 100% of the time, the fact that it A) only works up to 28mph, and B) only reduces the speed by 25mph, makes it inferior to other mfgr implementations.
Elon has stated that AEB will continue to mitigate crash energy at higher and higher speeds to lead the industry.
FlatSix911
Porsche 918 Hybrid
Latest update from Electrek ... Tesla Model S regains favor of Consumer Reports after Automatic Emergency Braking update
Consumer Reports has been able to grab a lot of headlines with its ratings on the Tesla Model S. Earlier this year, Tesla’s stock plummeted after they downgraded Model S because the new vehicles with the second generation Autopilot hardware didn’t have Automatic Emergency Braking. Now they are giving back the Model S’ top safety rating after Tesla updated the feature.
As we reported at the time, Tesla pushed Automatic Emergency Braking the same week that they downgraded the vehicle, but it was only enabled up to 28 mph. In June, Tesla updated the feature to work at full speed just like the first generation Autopilot and now over a month later, Consumer Reports says that it tested the feature and updated its rating:
“CR’s engineers then verified at our test track that the AEB operated at higher speeds by driving the electric Model S at a target that mimics the back of a car. We also verified that Tesla had sent the software update to its other cars by checking online forums and inspecting cars at Tesla dealerships.” What is more difficult to understand is why they didn’t test the feature in the first place since the publication originally gave the car the top rating even though it didn’t have AEB between October 2016 and April 2017 when CR had the full safety rating.
Similar threads
