Tesla autopilot HW3

[mongo]

Well, it could in theory do that for the forward three at least...

But again, in practice it does not work.

Once the NN is trained in drop (or other issue) detection, it can be propagated to the other cameras.
If you are referring to not working against light mist, that may be due to the defogger/ defrosted heaters. However, the dark night drop detection probably still needs improvement..

 

[electronblue]

Once the NN is trained in drop (or other issue) detection, it can be propagated to the other cameras.
If you are referring to not working against light mist, that may be due to the defogger/ defrosted heaters. However, the dark night drop detection probably still needs improvement..

No I’m talking about rain sensing working, which it doesn’t in any real sense.

 

[SpudLime]

No I’m talking about rain sensing working, which it doesn’t in any real sense.

The latest public update has some fixes for the wipers. Seems to be working much better for me. FWIW

 

[electronblue]

The latest public update has some fixes for the wipers. Seems to be working much better for me. FWIW

Haven’t received it yet but that FSD Early Access Program call is coming any minute now...

What concerns me most is rain sensing in low-light... without an active light source (like regular rain sensors) that seems trickier than necessary...

 

[SpudLime]

Haven’t received it yet but that FSD Early Access Program call is coming any minute now...

What concerns me most is rain sensing in low-light... without an active light source (like regular rain sensors) that seems trickier than necessary...

I havnt had much rain to verify, but ive seen other posts saying its better.

 

[emmz0r]

The latest public update has some fixes for the wipers. Seems to be working much better for me. FWIW

Still awful though. Almost never wipes when I actually want it to. 2019.4.3.

 

[SpudLime]

Still awful though. Almost never wipes when I actually want it to. 2019.4.3.

2019.5.15 might be better. Thats the one I had.

 

[mongo]

What concerns me most is rain sensing in low-light... without an active light source (like regular rain sensors) that seems trickier than necessary...

Well if you can tell there is water, there must be some light... (or they cheat and use microphone)

 

[rkarlberg]

There is no cherrypicking, I am looking at all deaths to which AP contributed, which AFAICT = 5 cases.

The statistics on deaths/injuries in other vehicles are immaterial, AP remains fundamentally unsafe at higher speeds and Tesla has shown no plan to make it safe. Even if tomorrow HW3 & FSD NNs rendered the vision system 99.99% accurate, in the case where the system fails to recognise from camera input that it is moving at 80mph into a static solid obstacle, the radar still provides no redundancy and another fatal crash can ensue if driver inattentive in the crucial moment. Then for actual L3 no driver attentiveness is required, making the fatal crash inevitable.

If the radar were upgraded to something fit for purpose, i.e. capable of reliably distinguishing stopped objects when moving at up to 90mph, L3 SAE may well be achievable without LiDAR. Certainly it is worth trying for.

Um, isn't that the very definition of cherrypicking - looking at just 5 cases out of 1,000,000? If your main concern is saving lives, wouldn't it make more sense to focus on reducing the 1,000,000 instead of reducing the 5?

Tesla is trying to reduce the total number of deaths, not to mention the other types of savings and convenience of FSD. I still just don't understand why some people want to bash Tesla so badly. There is no magic solution here; it's going to take time, money and hard work. But I want to wake up someday in a world where people can safely get to work, drunks can safely get home from a bar, blind/old people can get around... who else is doing more for that future than Elon Musk?

 

[Bet TSLA]

...of course accept that in dynamic situations AP works well to accurately detect vehicle motion ahead and thus prevents a good many crashes...

You know that ends the argument right there. You can't accuse Elon of being morally lax and yet agree that AP is quite possibly a net win. A few deaths can indeed be treated casually in the face of many saved lives. You may believe Elon isn't counting, but why are you so sure?

I'm pretty sure that it's totally obviously a win at this point. Yes, there are a few deaths caused by AP failures, but many more lives saved. The only calculus that makes sense is to put all effort into making AP solid enough to merit wide deployment, not in cleaning up unlikely edge cases. Save thousands of lives (and eliminate many more thousands of injuries) and worry later about the few extra deaths caused around the edges.

Your argument is equivalent to "Don't deploy any vaccines until we're absolutely sure they never cause any deaths." Plain stupid. Definitely immoral.

 

[electronblue]

Well if you can tell there is water, there must be some light... (or they cheat and use microphone)

My eyes can focus on the windshield to see the actual droplets. No Tesla camera is focused on the windshield so they are basically looking at shadows and shifts of light (like a rain sensor) which is harder to do without a light source lighting them up.

Not saying it can’t be done just seeing how it is hard. As for recognizing rain from the scenery, the resolution and processing power probably are not there yet either to have NNs output what weather they are seeing... and again that may be harder in low light...

There is a reason why Tesla’s AP2+ rain sensing tends to activate when coming out of a tunnel or when a car drivers towards you and its headlights light up the windshield...

 

[electronblue]

You know that ends the argument right there. You can't accuse Elon of being morally lax and yet agree that AP is quite possibly a net win. A few deaths can indeed be treated casually in the face of many saved lives. You may believe Elon isn't counting, but why are you so sure?

I'm pretty sure that it's totally obviously a win at this point. Yes, there are a few deaths caused by AP failures, but many more lives saved. The only calculus that makes sense is to put all effort into making AP solid enough to merit wide deployment, not in cleaning up unlikely edge cases. Save thousands of lives (and eliminate many more thousands of injuries) and worry later about the few extra deaths caused around the edges.

Your argument is equivalent to "Don't deploy any vaccines until we're absolutely sure they never cause any deaths." Plain stupid. Definitely immoral.

Your argument hinges on the belief that AP actually saves more lives than it risks. We really don’t know if especially Autosteer in reality saves more lives so it is a matter of belief. The argument changes if you don’t believe that.

 

[conman]

My eyes can focus on the windshield to see the actual droplets. No Tesla camera is focused on the windshield

Are you sure? Wide angle lenses can have a focal length of zero.

 

[Bet TSLA]

Your argument hinges on the belief that AP actually saves more lives than it risks. We really don’t know if especially Autosteer in reality saves more lives so it is a matter of belief. The argument changes if you don’t believe that.

Right. But Tesla has data, whereas the people (especially the one I was responding to) who are arguing here have absolutely nothing.

Elon has said before, quite explicitly, that getting AP widely deployed as soon as possible will save lives. It doesn't have to be perfect. So the argument being made here is that Elon is evil or stupid, as is his entire team. Neither is believable. It's obvious that he and his team believe that AP is saving lives as it is, and will save more as it gets better.

 

[SR22pilot]

I know you didn’t ask me but if it helps: I find Tesla as a company has been disappointing in many ways, leadership included. The car itself is good. Also AP2 as a product has been disappointing.

I don't have a problem with people being disappointed in things Tesla has done but the post I was responding to was way over the top. He assigned nasty motives to Musk.

I actually find AP a decent product. I think the full self driving has been over hyped with ridiculous dates. I think delivery quality could be better. Service has been good to me but I believe the horror stories. When I look at the issues I see a rapidly growing company with lots of growth paint. I also see a company in a grow or die situation that doesn't have the option to stop and fix things. I remember when my friends talked about the great service from Tesla but Tesla was selling a few thousand cars a year then and cars with six figure price tags.

 

[SR22pilot]

1. I can't actually know how he feels but am discussing the trade-offs he is obviously willing to make over a period of 3 years since the first fatality to which AP contributed. From this I can deduce that saving costs is more important to him than wasting a few more customers' lives.

2. He says AP is safer overall, because, well, he needs to keep selling it, but neither you nor I know whether he truly believes that. Certainly the stats he uses are pretty questionable.

3. I conclude he is cavalier about safety because he has failed to resolve a fundamental safety problem [AP ploughing into massive obstacles in planned path at 80mph without brakes or warning tone] yet now proposes to build much more complex functionality, likely to induce an even greater degree of driver complacency, upon this unsafe sensor basis, namely the inadequate radar and driver attentiveness monitoring system.

4. No, I've had relatively few problems with my car. Further I do not hate Tesla in the slightest and support the declared mission with much monies. What I am sceptical about is how the car I have paid for will ever reach the promised FSD in a manner which is safe and reliable, such that I could trust it to drive unsupervised at L3 without having to worry about my wife & children ending up plastered @80mph into the tail-end of the first stationary traffic jam it encounters?

5. IMHO the only way that can ever happen is with, at minimum, a radar sensor upgrade to go along with HW3. And a DAMS worth a damn, i.e. which tracks that the driver's eyes are actually open and focussed on the road ahead, would also be a great help in getting to L3 approval without unnecessarily wasting further customers/3rd parties.

Well, let's start.

1: Neither of us have any idea what has been done internally. You see takeoffs knowing nothing of the internal discussion or issues. Are you aware that the issue with stopped vehicles is present on pretty much all other brands and even called out in several manuals.

2: I too have questions about the statistics. I personally think he believes the stats but I have no more reason to believe that than you do to believe he doesn't.

3: Look at the Tesla crash data. Is this a company cavalier about safety? You say he is cavalier about an issue because he hasn't resolved an issue present o other cars. You aren't calling them out. One of the things that irks me most is the double standard people have towards Tesla. If a Tesla car catches fire it is national news. If other cars catch fire nothing is said. If a Tesla is in a wreck on AP it is nationally reported. If another car is in a wreck while on cruise control people probably don't even know because the car doesn't report it. In some countries trucks have to have side skirts to decrease accidents and aid automatic detection. Hav you looked at the problem of radar detecting a stopped car form a rise in the highway? Have you seen that Infiniti, as an example, lists stopped vehicles as an issue. Are you yelling in public that those CEO's must be cavalier about lives because they haven't solved the problem?

4: Here you have the attitude that I fear will cause tens of thousands to die. We will have FSD capabilities that will be safer than people driving themselves but they won't be used because every FSD accident will be national news. I doubt FSD will be perfect until humans aren't allowed to drive. Today a car pulled across in front of me. My car beeped loudly. I use AP when I am in stop and go traffic and I find it very useful. I feel it is safer than I am since I could easily miss taillights going on after 45 minutes of traffic torture. I find that NoA is good at making sure I take the correct exit but I keep a very watchful eye on it. One accident that happened had given a guy so much trouble he had reported it to Tesla something like 7 times. I would have been hyper vigilant. A lady hit a stopped firetruck because it was uncovered by a moving car. This is a huge problem for all systems. I think she should have sent Tesla a thank you letter since all she got for texting and driving into a 60 mph crash was a broken ankle.

5: I suggest you look deeper into the technical issues involved. I think side skirts for trucks that are radar reflective are a change needed now. Where is your outrage over that? I have seen nothing that says Tesla's radar system is inferior to that in other cars.

Now for a case you won't mention. I see articles about drunks asleep in Teslas and people seem horrified. I am grateful. I am grateful that AP kept the car from hitting any other cars. I am grateful that because of AP the police were able to stop the car in a controlled manner. DO you think drunk people only drive Teslas? Are Teslas the only car people fall asleep in at the wheel? MY girlfriend's dad died because he fell asleep at the wheel. Had he been in a Tesla on AP he would probably be alive. It was just a brief moment according to what the doctor thinks happened.

 

[Bladerskb]

I'm pretty sure that it's totally obviously a win at this point. Yes, there are a few deaths caused by AP failures, but many more lives saved.

Theres absolutely no proof of this. Infact there is proof of the opposite. There have not been a single death in an ADAS system other than Tesla AP. That is huge!

 

[boonedocks]

Theres absolutely no proof of this. Infact there is proof of the opposite. There have not been a single death in an ADAS system other than Tesla AP. That is huge!

Tho you “might” be correct I don’t know that I would assume you are. For whatever reason if a Tesla gets a flat tire it makes the news (stretching that a bit but totally true). If Cadillac or BMW or Audi had an ADAS related death I am not certain it would make tonight’s World News Report

 

[SR22pilot]

1. No, I have no special insight on that but of course accept that in dynamic situations AP works well to accurately detect vehicle motion ahead and thus prevents a good many crashes, which are perhaps in the main low speed fender-benders in dense traffic.

2. That laudable fact however does not compensate for the treacherous weakness that at any moment doing 80mph on the highway it is liable to go full tilt into a massive obstacle parked in the planned path, the type of accident on restricted-access highway with the highest probability of resulting death.

3. It has neither been explained by Tesla why this weakness persists after 3 years, nor how they plan to eliminate it before building FSD upon the same sensor basis.

From a 2019 Cadillac owner's manual:

ACC may not detect and react to stopped or slow-moving vehicles ahead of you. For example, the system may not brake for a vehicle it has never detected moving. This can occur in stop-and-go traffic or when a vehicle suddenly appears due to a vehicle ahead changing lanes. Your vehicle may not stop and could cause a crash.


From a Car & Driver article:

The edge cases cover the gamut from common to complex. Volvo's owner's manuals outline a target-switching problem for adaptive cruise control (ACC), the convenience feature that relies on the same sensors as AEB. In these scenarios, a vehicle just ahead of the Volvo takes an exit or makes a lane change to reveal a stationary vehicle in the Volvo's path. If traveling above 20 mph, the Volvo will not decelerate, according to its maker.

So Musk is somehow a cold uncaring person because he hasn't solved an issue that other cars also have.

 

[SR22pilot]

There is no cherrypicking, I am looking at all deaths to which AP contributed, which AFAICT = 5 cases.

The statistics on deaths/injuries in other vehicles are immaterial, AP remains fundamentally unsafe at higher speeds and Tesla has shown no plan to make it safe. Even if tomorrow HW3 & FSD NNs rendered the vision system 99.99% accurate, in the case where the system fails to recognise from camera input that it is moving at 80mph into a static solid obstacle, the radar still provides no redundancy and another fatal crash can ensue if driver inattentive in the crucial moment. Then for actual L3 no driver attentiveness is required, making the fatal crash inevitable.

If the radar were upgraded to something fit for purpose, i.e. capable of reliably distinguishing stopped objects when moving at up to 90mph, L3 SAE may well be achievable without LiDAR. Certainly it is worth trying for.

Other statistics are immaterial? How ridiculous. You are setting up a case to crucify Tesla if there is one death and exonerating millions of other deaths. I bet those deaths mitered to their loved ones. If AP decreases deaths then I say it is a positive improvement even if it isn't perfect. You citing 5 deaths is not statistically significant to prove it is more dangerous than a person driving. Using your logic we should eliminate all airbags because in some cases they may have caused rather than prevented a death. How about the old "thrown clear of the accident" scenario used by people to argue against seatbelts. It seems you fall into that category. Seatbelts aren't perfect so we should eliminate them.

If you want to provide data showing AP is dangerous compared to a person driving then please do that. So far you haven't. You have just made accusations.