You need to read how AP works. The system can and should be blamed when it fails to operate as designed which isn't uncommon.
My goodness. Is Tesla being purposefully misleading/dishonest here? Or is it merely careless wording? Hands-not-detected-on-wheel does NOT mean hands-off-wheel.
So.... Are You saying since she kept altering Autopliot settings and speed, she is not familiar with the system? And if she's not familiar with it, that's why the accident happened? Does that mean Tesla warns that Autopilot accidents happen with inexperienced owners?
I am disappointed with this response. "Hands not detected" is silly. There is no way for the car to know if your hands are on the wheel, unless you are giving a torque to it in any one direction, which obviously we will not do it. So i am sorry, "Hands not detected" is bogus and is not get out of jail card for Tesla. The proper explanation would be, "Tesla's cruise control will not detect stationery objects (that were not in motion prior) above 40 mph which is consistent with any manufacturers intelligent cruise control at this time. Being observant and using AP as a driver's aid is what Tesla recommends, which in this case the driver did not adhere to". They are going to receive a lot of flak for this explanation. NHTSA might simply recommend to NTSB to suspend AP.
Perplexing for sure. 40000 miles on AP1 and it has never failed to slow down/stop for a vehicle in front of me. I don't get this rear ending of fire trucks with AP activated...is this just an AP2 thing? Maybe someone(I'm too stupid/lazy) could create a poll asking: 1. If AP has ever failed to slow down/stop for a vehicle and 2. was AP1 or AP2 being used
Thanks, read the article...but it doesn't really explain why it doesn't work. There is a comment that says: "The radars they use are apparently meant for detecting moving objects"...is this correct? why doesn't work with stationary object? maybe is the bounce of the signal not right and not received? is a Doppler effect issue? Lack of processing power? are there better radars in the market capable of detecting stationary objects? can we make stopped emergency vehicles more visible to radar? What about teslavision, why it failed? There are so many questions and blaming the driver is not cutting it for me.
Why does the report strangely scroll back to the top? I can only read a few sentences before it does that...
This vehicle was a AP1 car, I think that the Model X in Mountain View was as well, is this an issue with the mobileye Hardware being limited vs the newer AP2+ hardware? I have no information about this just curious if the folks on this forum that are more up on the differences would have details. The crash in Florida by the kids that were speeding was also an AP1 car from what I understand.
Radars get a lot of 'ghost images' that are artifacts. These ghosts are a real problem. So a lot of filtering goes on including filtering out stationary objects. I think other sensors have the same problem. If camera algorithms are the same as what some military rags talk about, there is something called change detection. If the object is not moving, it is not changing...
IMHO, the problem might be related to the processing speed of what the computer sees as a problem. I don't think it is fast enough. Also the scenario of a stationary vehicle ahead of a vehicle you are following is arguably a daily occurrence during rush hour when traffic porpoises and cars go from 50 to 0 and back up again, while everyone is changing lanes trying to get that one car advantage that will get them to their destination 30 seconds earlier. BetaPilot is not ready for the real world!
All of the above! The primary function for Doppler RADAR is to NOT detect stationary objects. It's job is to detect anything that registers an above zero mile per hour which is understood as a moving object. Street lights, overhead signs don't register a speed so they are ignored. That's a quite neat trick as long as those stationary objects are not in the middle of the street such as the fire truck in this case!
Thank you! I have the same complaint; I hold the wheel and often get a written warning - I hope Tesla doesn't have me on a dodgy drivers list! In my i3 which often flipped from cruise to max regen if it saw a squirrel, the sun or a bridge (it's a very nervous TACC) I subconsciously learnt to react to sudden deceleration and loud chimes by quickly pressing the accelerator...firmly... I know, yeah, nuts right?!? However, I'm learning that to drive my Tesla I have to sway left-right-left all the time like I'm in some 60's in-car movie scene... I sometimes switch Auto-steer and TACC off because, I mean, why bother?
Having a technical background I would agree with you. What I am worried about is the legal system in the USA. It seems to me that in the USA, when people get hurt or hurt someone due to a product that they bought and used in a manner that was incorrect, yet demonstrably common, then there is a real risk that the manufacturer of that product will be held liable for damages that this incorrect yet common usage caused. So I think that Tesla is at risk of ending up in a AP-abuse lawsuit that they may or may not win, but which even in the best case can be a drain on their resources and on their public image. Unlike the risks associated with getting the M3 production ramped up, I am very unsure that Tesla has the risk of legal fall-out from incorrect but common AP-abuse under control. If anyone can explain why the above worry is unfounded, then please go ahead. Btw, why is AP-abuse almost(*) always described (also by Tesla) as incorrect, as opposed to illegal? The (traffic) law supersedes any product manual or guideline, so I would consider it natural to point out that AP-abuse is simply breaking the law, since the driver at all times must be attentive and keep at least one hand on the steering wheel (at least for now, in sane jurisdictions). (*) There was that one UK Tesla-driver who had his driving license suspended for engaging AP and then moving to the passenger seat. A serious contender for the Darwin-awards.
That is a very good point Iklundin. I agree 100%. This really is the bottom line isn't it. My understanding which is consistent with all of my Tesla Autopilot driving experiences with respect to the manner in which Tesla's Autosteer assistance technology operates is that some minimum amount of steering wheel torque (how much?) applied without actually deactivating Autopilot continuously resets the Autopilot warning timer. I can understand why this operational implementation might make the driver unsure if the Autosteer system is operating correctly but not really if you examine in detail the Tesla Autopilot activation warning initially presented to the driver. In my opinion I think Tesla could make the warning message more effective by presenting it in a larger font (2x the present size) and for a longer period of time, e.g., 15 seconds instead of 10 seconds.
I am not saying that you are wrong either, but is sure sounds like you are talking about using autopilot for airliners in Class A airspace (over 18,000 ft.) That is one specific application. I fly far below that and in my aircraft, the flying pilot is ALWAYS responsible for watching outside the aircraft with hands on the controls and ready to take over in a seconds notice. Just like autopilot in a Tesla. Some people will use any excuse to be on their cell phones while driving. It doesn't matter what Autopilot is called. People do the same thing or worse without Autopilot. Personally, I find myself more aware of my surroundings while using Autopilot,.
Actually, I'm quite familiar with how radar works and see no reason whatsoever the car couldn't have detected the object in front of it. Do you have any particular reason for thinking radar couldn't?
It just takes a tiny tap to get the steering wheel to notice you. I mean, go ahead and play Bullitt if you want, but it's not necessary just to keep AP engaged.