Tesla did make changes to their autopilot system including increasing the “wheel nag“ interval, they just didnt respond to the NTSB, according to that guy.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
U.S. opens formal safety probe for Autopilot - 2021 Aug 16
- Thread starter Terminator857
- Start date
I'm not sure why we think "Tesla Vision" will solve the OEDR static object etc issues behind these crashes when Tesla themselves said in November 2020 that they don't expect any significant enhancements to the OEDR. These systems will continue being Level 2, even a final release of City Streets will be Level 2, and I think the NHTSA will strictly be looking to build in more robust driver attentiveness monitoring systems to ensure users are always ready to take over and navigate around stuff the car can't handle.
From reading Tesla's letter to the DMV, these systems won't even be able to give us any warning of crashes similar to the incidents being investigated by the NHTSA. Drivers legitimately need to be able to take over in a split second 100% of the time.
From reading Tesla's letter to the DMV, these systems won't even be able to give us any warning of crashes similar to the incidents being investigated by the NHTSA. Drivers legitimately need to be able to take over in a split second 100% of the time.
I'm not sure which model you were referencing in particular, but the model 3's stopping distance is about 133ft, (can be less depending on the tires). I believe the M3P is a lot less. So, it's closer to 40m or under from 60-0.
I used the numbers I found at link below. The point stands anyway, margins are too small when speed is high. I would love to see some accurate numbers and calculations.
The Importance of Correct Stopping and Following Distances
Rear end collisions are the most common accident type in Australia, so it is important to understand the correct stopping and following distances.
ubicar.com.au
mwfriederichs
Member
It certainly does. A few weeks ago, I was slowly (less than 15 mph) passing an accident scene, when a police officer stepped out from between two parked cars towards the lane I was in and my Model 3 SLAMMED on the brakes. Tires squeaked and police looked up and over at me, and waved me on.
Different agencies. NHTSA is the one with teeth to enforce policy, NTSB does not. NHTSA is also more influenced by manufacturers lobbying when making their rules, though. So it may depend on how much or well Tesla can lobby them.
IS this a totally separate investigation?
www.cnn.com
Senators call for federal probe into Tesla's Autopilot claims
Tesla's Autopilot feature could soon face its second federal probe of the month.
www.cnn.com
Yes, this is about Tesla basically lying to customers about the FSD claims.IS this a totally separate investigation?
Senators call for federal probe into Tesla's Autopilot claims
Tesla's Autopilot feature could soon face its second federal probe of the month.
drdumont
Member
And two vehicles cannot occupy the same space at the same time, without something getting damaged.
Nasty thing, Physics... Lots of those laws really are hard to break without dire consequences.
However, comma, I will likely be sitting in the Old Engineers' Home Gumming my Jell-O and wishing for The Good
Old Days long before truly autonomous vehicles are commpnplace.
Denks Gott.
I am sure there will be more accidents and deaths without using AP on the highway. Although AP is not in L5 level yet, but it somehow can drive safer by using the 360 degrees "eyes" and sensors to detect more up-coming danger than what a human eyes is capable. With trillions and trillions dollars has invested in AI industry worldwide. I believe L5 AP is coming soon in the near future. XD
This is no investigation at all. It is just sen (ihatetesla)Markey writing a letter complaining about technology he doesn't understand.IS this a totally separate investigation?
Senators call for federal probe into Tesla's Autopilot claims
Tesla's Autopilot feature could soon face its second federal probe of the month.
He also introduced a bill to keep the kids off his lawn.
My feeling about all of this is simple: If AP was used as recommended, with the driver paying attention and taking control when necessary, the majority of these accidents would not have happened.
Yes, even I had unrealistic expectations of Autopilot / FSD when I bought the car because of the hype, however, it quickly became apparent, through practice and from reading the abundant warnings, that the system required active monitoring by the driver. It didn't take a rocket scientist to figure that out. Whatever hype was out there no longer mattered.
The fact that people could so easily throw common sense out the window (assuming they had any to begin with), and carry on as if the system was fully autonomous, is bewildering and an eye opener.
We could talk about Elon and his misleading statements in his overhyping of AP's capabilities, and I would have no argument. However, reality and common sense should take precedence over the hype.
And while Tesla's (and Elon's) feet should be held to the fire over AP's actual capabilities, the Senate investigation is looking more like a witch hunt than anything else.
Yes, even I had unrealistic expectations of Autopilot / FSD when I bought the car because of the hype, however, it quickly became apparent, through practice and from reading the abundant warnings, that the system required active monitoring by the driver. It didn't take a rocket scientist to figure that out. Whatever hype was out there no longer mattered.
The fact that people could so easily throw common sense out the window (assuming they had any to begin with), and carry on as if the system was fully autonomous, is bewildering and an eye opener.
We could talk about Elon and his misleading statements in his overhyping of AP's capabilities, and I would have no argument. However, reality and common sense should take precedence over the hype.
And while Tesla's (and Elon's) feet should be held to the fire over AP's actual capabilities, the Senate investigation is looking more like a witch hunt than anything else.
stopcrazypp
Well-Known Member
The fact of the matter is there is no evidence this played any factor in the accidents that have happened. As an owner, as soon as you experience that first disengagement, any illusion about it driving itself is gone regardless of any hype. There's only a very narrow window where a driver that have never experienced a disengagement may be misled and get into an accident. The rest are people who may be doing it for a stunt like that Tik Tok guy who was doing that backseat stunt that got the media all up in arms about Tesla's driver monitoring. C&D just did a recent article that showed that the other manufacturer's driver monitoring was pretty much the same in terms of being able to be defeated, with some even more easily defeated.
Consumer reports needs to re evaluate there driver assist ratings since Tesla outscored everyone on every metric except driver engagement. If all of the systems can be tricked if day they are all equal in that respect
Autopilot lost in three categories in the Consumer Reports tests: Keeping the Driver Engaged, Clear When Safe to Use, and Unresponsive Driver.
If you believe the driver engagement metric should be binary and leave zero room for ranking of the systems by difficulty in defeating them, well I doubt many would agree. A system that can be defeated by a simple weight attached to the steering wheel should very clearly be ranked lower than a system that is defeated like this
Consumer Reports used a 1-10 ranking system and that certainly feels appropriate
All of the other systems were easily defeated to though. Any system they can be easily defeated shouldn’t score higher than a 1 or a 2 on driver engagement IMO.Autopilot lost in three categories in the Consumer Reports tests: Keeping the Driver Engaged, Clear When Safe to Use, and Unresponsive Driver.
If you believe the driver engagement metric should be binary and leave zero room for ranking of the systems by difficulty in defeating them, well I doubt many would agree. A system that can be defeated by a simple weight attached to the steering wheel should very clearly be ranked lower than a system that is defeated like this
Consumer Reports used a 1-10 ranking system and that certainly feels appropriate
Unresponsive driver and driver engaged seem like the same test to me. I will agree that Elon’s propaganda costs them points on clear when safe to use, although even a casual flip through the owners manual will clear that up.
Driver Engagement is not the same as Unresponsive Driver, and having a system that uses steering wheel torque to measure engagement is missing the whole eyes-on-the-road aspect. I can sit in a driver's seat and put torque on the wheel while looking down at my phone, that doesn't mean I'm at all engaged in what's happening in front of the car.
You're reaching hard and disregarding a lot of nuance. Most of the systems tested by Consumer Reports did score within one point of each other in Driver Engagement, then SuperCruise obliterated everyone and for good reason. Monitoring driver's eyes is clearly the best solution out there right now, and there is likely little risk of people wearing eyeball glasses to defeat it if only because that would make other activities (like looking at your phone while the system does its thing) more difficult than it would be worth.
This is like arguing that all locks should be trashed because all locks can be defeated. Everything can be defeated with enough effort and the correct tools, but the point is in making it onerous enough that people won’t bother doing it.
Last edited by a moderator:
I will claim defeating is irrelevant in other brands. At least the ones I have owned. I-PACE, e-tron and ID.4 all will disengage their AP equivalent without warning sounds if the system feels for it. This only scare you once and after that you definitely keep eyes on the road and hands on wheel ever since. Any sane person will never think of trying to defeat those systems because any lost road marking will possibly, not always, disengage the system.
Tesla AP1, 2 and 3 though, all iterations will try to keep on even if the system is confused by on/off ramps, tight bends etc
Something a safety probe might look at is whether the information displays the driving mode clearly.
For example, say you change drivers and your friend changes all your settings, response and following levels, creep mode, etc. Or this happens while you are out of the car, or perhaps you changed some settings hours ago and just forgot you did it.
Are all the relevant modes displayed all the time? If you expect something to be active but in fact it's turned off - is there any safety component to that problem?
If the car cancels an Auto feature and you miss the message or the radio/wind noise obscures the beeps is it clear that you are no longer protected by that Auto feature? Are there large and obvious mode indicators?
I'm not saying Tesla isn't good, but is it the best, or are there better ways of informing the driver what the current settings are - without having to wade into UI menus to discover what the settings may be.
For example, say you change drivers and your friend changes all your settings, response and following levels, creep mode, etc. Or this happens while you are out of the car, or perhaps you changed some settings hours ago and just forgot you did it.
Are all the relevant modes displayed all the time? If you expect something to be active but in fact it's turned off - is there any safety component to that problem?
If the car cancels an Auto feature and you miss the message or the radio/wind noise obscures the beeps is it clear that you are no longer protected by that Auto feature? Are there large and obvious mode indicators?
I'm not saying Tesla isn't good, but is it the best, or are there better ways of informing the driver what the current settings are - without having to wade into UI menus to discover what the settings may be.
Scene image variability and complexity have always posed a challenge to image processing and understanding experts. I know, I worked that technology for autonomous platforms and hats down to the imaging team at Tesla but FSD will never be 100% Full Self, it will be AFSD - Almost Full Self driving and driver interaction will always be required. That said, I am surprised that ONLY such a small number of incidents have been reported !
Similar threads
