Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Tesla has allegedly activated a selfie cam Driver Monitoring System

This site may earn commission on affiliate links.
@linux-works

Your statement that you work in the software industry is not surprising, given your user name here. The fact that you both work in the software industry, and in the particular field you are complaining about the actions in, and one of the reasons you bought a tesla was to "eat the dogfood" (IT parlance for "walk in the shoes of your end users") makes more sense.

You disagree with the design choices, because you are in that field and would not have done it that way. Nevertheless, your assertion that "the wheel tug works" has been proven to be wrong. It is easily defeated, with multiple people posting videos of how to do that, and the bad publicity it brings.

Its my belief that the internal camera in the rear view mirror was NEVER intended for "driver attentiveness". Its in the wrong position for that. It was even stated by elon that the camera was for monitoring the inside of the vehicle as a robotaxi. It IS in a good position to keep an eye on the back of the car, but not in a great position to keep an eye on the driver.

I feel this is in relation to the bad press from media (both online and others) because stupid idiots posted videos of riding in the back seat, and some people also appear to have been doing that when they died somewhat recently. That brings regulatory eyes, and THAT (oversight) brings "oh #@@ what do we do to get these regulators off our backs".

So, just like they created sentry mode because of break ins in San Fran, they activated the camera to get regulators off their backs about "what are you doing to prevent people from getting out of the drivers seat, because what you are doing now DOESNT WORK".

So, no, "wheel nag" doesnt work, thats proven and indisputable. Peoples lives have been lost because it doesnt work.

Now, wanting to be able to "not apply updates" is actually something I agree with, IF the person also is willing to go with the consequences of that action, which in this case likely is the disablement of autopilot / FSD. Someone should be able to push a button and "opt out" of the camera, and decline FSD.

Your argument of "I bought it this way, and it should stay this way if I want it to" is valid for a self contained system, but not for a "software as a service" type system, but you work in the industry and likely know that better than I do. I work in IT, but am not a developer so your knowledge around this would be much greater than mine or most of us, really.

==================

TL ; DR --

you are complaining about this because its a design decision you wouldnt have made and dont agree with, and you are in the same industry. Your assertion that "the nag works" is incorrect, and has been proven to be so multiple times, with some people apparently losing their lives because they believed the car could do stuff it doesnt. Your desire is a fair one, in my opinion, if one is also willing to decline using FSD. My opinion is that the camera was never intended to be used for this, but is being turned on due to regulatory pressure.
 
I'd like to understand why people think the nag and wheel tug *has* to be replaced and why some consider it a failure?

this touches on how much nannying you want the car to take on. and there is a line, not always bright, between where the vendor did due dilligence and decided on a balance between being very strict and being very loose. somewhere in there is the right value and because we are all doing R/D as we go along, there are no set answers. we're all eating the dogfood and that's fine, but we're also learning from all this experience and can refine the design decisions. that's what 'agile development' is essentially about.

before I bought my tesla, I didn't know if the nags would annoy me. now, after more than a year of ownership, I'm fine with it, especially since I can just click up or down the volume scroll wheel and not even have to move the steering wheel itself. I like that even more. no reason to have to have multiple states on the steering. tug just a little to show driver is alive, tug more to actually go into 'direction change' mode and affect steering. its ok, but I like the volume rotary knob idea much much better and would really miss it if that goes away.

I think that tesla made a good break-point in where their due dilligence stops and the driver has to accept responsibility for the end result of the car's motion. I'm quite ok with level 2 and when higher levels are truly do-able, the new hardware will likely be needed and maybe by then, the software systems will be more industry standard. maybe v2x will finally catch on, as well. I hope it does, as its a super great idea, on paper, at least.

I don't want or need the car to nanny me. yes, the wheel tugs can be worked around. that's not at all a 'proof' that the design or implementation was a failure. the optical system will also have false pos and neg results and people will die. this is a given that you accept when you make things that can cause people to get hurt. you can't be 100%. but you do the best you can and keep the balance in mind.

I'll tell you this: in china, the telemetry goes to multiple government bodies, many more times than once every 10 seconds. I really dont want to see the US get anywhere near that bad; and I dont ever really want cameras in my car that I didnt put there, myself.

options are good. some want the cameras. but not all do, and its intellectually dishonest to call the wheel tugs a failure. its not a failure. its a style that can be abused, but all systems can be abused if you try to defeat them. I dont want the vendor trying to solve the micro-problems (the 'wheel defeat beanbag people' are a tiny % of a tiny %. not a real world problem.)
 
I'd like to understand why people think the nag and wheel tug *has* to be replaced and why some consider it a failure?

this touches on how much nannying you want the car to take on. and there is a line, not always bright, between where the vendor did due dilligence and decided on a balance between being very strict and being very loose. somewhere in there is the right value and because we are all doing R/D as we go along, there are no set answers. we're all eating the dogfood and that's fine, but we're also learning from all this experience and can refine the design decisions. that's what 'agile development' is essentially about.

before I bought my tesla, I didn't know if the nags would annoy me. now, after more than a year of ownership, I'm fine with it, especially since I can just click up or down the volume scroll wheel and not even have to move the steering wheel itself. I like that even more. no reason to have to have multiple states on the steering. tug just a little to show driver is alive, tug more to actually go into 'direction change' mode and affect steering. its ok, but I like the volume rotary knob idea much much better and would really miss it if that goes away.

I think that tesla made a good break-point in where their due dilligence stops and the driver has to accept responsibility for the end result of the car's motion. I'm quite ok with level 2 and when higher levels are truly do-able, the new hardware will likely be needed and maybe by then, the software systems will be more industry standard. maybe v2x will finally catch on, as well. I hope it does, as its a super great idea, on paper, at least.

I don't want or need the car to nanny me. yes, the wheel tugs can be worked around. that's not at all a 'proof' that the design or implementation was a failure. the optical system will also have false pos and neg results and people will die. this is a given that you accept when you make things that can cause people to get hurt. you can't be 100%. but you do the best you can and keep the balance in mind.

I'll tell you this: in china, the telemetry goes to multiple government bodies, many more times than once every 10 seconds. I really dont want to see the US get anywhere near that bad; and I dont ever really want cameras in my car that I didnt put there, myself.

options are good. some want the cameras. but not all do, and its intellectually dishonest to call the wheel tugs a failure. its not a failure. its a style that can be abused, but all systems can be abused if you try to defeat them. I dont want the vendor trying to solve the micro-problems (the 'wheel defeat beanbag people' are a tiny % of a tiny %. not a real world problem.)

The bad publicity that comes from people dying from defeating it show its a failure, and can impact tesla financially, which is what matters here in the US. The bad press brings increased regulatory attention, and tesla already receives more of that than most other companies.

its not intellectually dishonest to call the wheel tug a failure, Its fact. Its been proven as not effective, too easy to be defeated. Not much different than codes when building homes that prevent people from doing X or Y, those codes come about because people did X or Y even though "they shouldnt" and people got hurt or died.

Now that it has been proven without a shadow of a doubt to be easily defeated, tesla has a duty to do "something else" to enforce "stay your Q@RR@ in the seat". I dont like cameras in the car either, for you to keep stating "the nag works" when its clear it doesnt, AND amid proof that people will continue to exploit it to try to get their "15 minutes of fame" because people are stupid, is confusing.

There are traction control systems on vehicles, because people will push the accelerator all the way to the floor, even if its not safe, and they have to be protected from themselves. This is not much different than a traction control system (that people still try to defeat).

With that being said, I am not one who enjoys "online debates". I have my opinion, of course you have yours. I dont think this is going to change for tesla (or any other company that pushes updates as part of the OS, to control it) unless they are forced to by legislation, and I dont think thats going to happen any time soon.

These are not technical limitations, they are design choices. I dont see companies going toward "less control" because they all want more, not less control.

In any case, I will say, i hope you come to a decision that works for you, whatever that is" and bow out of this discussion.
 
Last edited:
  • Like
Reactions: rxlawdude
time will tell how much the optical system can be tricked and if those same percent of 'bean baggers' will do the same once tesla switches over to requiring the camera.

even just the driver wearing sunglasses, I'm not sure how an optical system can *ensure* to any greater ability than the tugs, that the driver is alert and able to take over when prompted to.

as for whether you like online debates, that's not for me to say, but you are certainly free to stop posting, yourself, if this subject annoys you. I think user debate on how the various systems work and don't work, and how they annoy and dont annoy the users is right on topic for this thread. tesla reads this site and so its useful for them to hear direct customer feedback, when given with rationale and backing justifications. I've given mine and I hope tesla takes them to heart.

we all want the car to be better, overall, and we want the most acceptance for this new tech so that it continues to be adopted by the masses.

insisting on optical sensors just to ensure level-2 - I think that goes too far and tesla needs to hear that message from the users who feel strongly about it.

if they only require the camera for use beyond level-2, I think that would be a great compromise. don't you agree? give the option to stay at level 2 camera-less (interior)?
 
So how come Tesla is not activating the driver facing camera for driver attention immediately on all Model 3 and Y's? Why only on new cars first that are using Tesla Vision? IMO, it would seem like something that would be good to push out the entire fleet of M3 and MY's ASAP.
 
  • Like
Reactions: jsmay311
My wife's 2021 Subaru has a driver monitor that beeps at her whenever she looks over at a neighbor's house while driving through the neighborhood. I love it! :p

I can't for the life of me understand why someone is concerned about a camera pointed at them for this purpose. Like your life is so interesting? Please.

The scroll wheel substitution for the steering wheel shimmy only works so many times, then the system will ignore those and you have to shake the wheel. I can get maybe 30 minutes adjusting volume on AP, until it won't accept that. I'd gladly take a camera system over the wheel shake.
 
So how come Tesla is not activating the driver facing camera for driver attention immediately on all Model 3 and Y's? Why only on new cars first that are using Tesla Vision? IMO, it would seem like something that would be good to push out the entire fleet of M3 and MY's ASAP.
Newer camera with higher resolution ?

ps : The other thing is "change management". Easier to put new features in new cars (with new drivers) than change something that people are used to.
 
The bad publicity that comes from people dying from defeating it show its a failure, and can impact tesla financially, which is what matters here in the US. The bad press brings increased regulatory attention, and tesla already receives more of that than most other companies.

Elon brings a lot of the attention and scrutiny by his actions. And this isn't about his hinting that FSD will be here any day now for 6 years. He likes to brag to the
world that the (skewed) data proves his vehicles are x-times safer. Safest car ever made, etc etc. He seeks that extra attention(but of course doesn't like it when the attention isn't fan-boy worship)

But I'm not saying what everyone here doesn't already know.
 
  • Like
Reactions: Tim42069 and Darmie
The bad publicity that comes from people dying from defeating it show its a failure, and can impact tesla financially, which is what matters here in the US. The bad press brings increased regulatory attention, and tesla already receives more of that than most other companies.

its not intellectually dishonest to call the wheel tug a failure, Its fact. Its been proven as not effective, too easy to be defeated. Not much different than codes when building homes that prevent people from doing X or Y, those codes come about because people did X or Y even though "they shouldnt" and people got hurt or died.

Now that it has been proven without a shadow of a doubt to be easily defeated, tesla has a duty to do "something else" to enforce "stay your Q@RR@ in the seat". I dont like cameras in the car either, for you to keep stating "the nag works" when its clear it doesnt, AND amid proof that people will continue to exploit it to try to get their "15 minutes of fame" because people are stupid, is confusing.

Tesla doesn't have a duty to do anything until forced by law.

There are hundreds of millions of vehicles on the road that were found to have previously unidentified weaknesses with collisions.
Newer vehicles have those improvements, older vehicles are not modified.

The software can be treated exactly the same way, especially given that there are 2 very different code paths for the camera and the wheel activity.

Besides, this is hypothetical. The poster can wait and see.
 
I am all-in regarding safe deployment of new technologies, but I think there are some fundamental precepts that are unfairly overlooked regarding the intentional hacking of AP systems. Let's step back and consider automotive technology we're all pretty comfortable with, that YouTubers e.g. Consumer Reports could easily "expose" and train us to kill with, yet they don't because it would elicit yawns if not ridicule:

If cars traditionally had only single-pedal control (press to accelerate / release to brake), then a design with a separate brake pedal would be deemed inherently and insanely dangerous.

If cars traditionally engaged an auto-brake while stopped / idling, then a design that creeps forward by itself at stoplights would be deemed inherently and insanely dangerous.

If cars had traditionally been powered by giant clock-springs or compressed-air tanks and only achieved 10-mile range, then a 400-mile design that explosively combusted volatile and toxic liquid fuel, pumped by the gallon into an onboard holding tank by uncertified civilian operators, would be deemed inherently and insanely dangerous.

This could be continued and, of course, easily extended beyond automobiles to kitchen knives, appliances, lawn and shop tools, really almost everything to some degree.

The precepts that guide us out of a cowering and nanny-filled existence (and therefore out of a controlled and unfree existence) are:
  • Children should be looked after, to a reasonable but not suffocating extent, by responsible adults.
  • Adults are expected to look after themselves and their / the community's children.
  • We should want and expect children to grow up and become responsible adults. If they don't then we hold them primarily accountable, not the manufacturer of their chosen implement of irresponsibility.
  • To this end, products should be applied to their intended purpose and should be reasonably safe & predictable, yet effective in operation. Same goes for food and medicine.
  • Accidents happen but we endeavor to learn how to reduce them. This however involves responsible use as well as responsible design & construction of products. Accidents from confused or unintentional misuse are top-priority learnings; predictably bad outcomes from intentional abuse are life teachings.
  • Technical progress involves risk but is deemed good when it tends toward reduction of risk, cost and stress, and towards expansion of health, wealth, happiness and freedom.
I expect Tesla and its competitors to take reasonable steps to make emerging technologies safe, compatible with current roadways and easy to use correctly - and hopefully affordable in the process. It's unreasonable to expect them to counteract imaginatively malicious abuse. Yet, such unreasonable expectations, and the fickle evaluation of known/pervasive/accepted risks vs. unfamiliar/unlikely/novel risks, are at the core of so many autonomy discussions, including the now-predictable pattern of sensational but wrong and misleading news coverage.
 
  • Like
Reactions: Ruffles
Two things that could be relevant here:
  1. Doesn't the description of the DMS specifically say the video never leaves your vehicle? And so one might say "yeah, but I have to trust them" - well, the car also has a microphone that only records when you push the scroll wheel. This is the same category, we have to trust they aren't recording all the time. Or storing/selling your location data, etc.
  2. I thought somewhere in the T&Cs of the warranty, you are "required" to accept software updates. Yep found it (p10):
1622390471193.png

Not arguing against a preference that you can de-activate it if you want to.
 
Green has done some testing of the new camera driver monitoring.

That's interesting and I think generally good news in how they intend to use this. Personally, given how each version fails differently and the general rate of change (each new version allows new scenarios to be solved correctly but new modes of failure) i personally feel the hand on the steering wheel is safer. The reason being you can immediately feel the car not doing what you expect. This is not the same as just paying attention. Some seem to feel it is a huge pita, but if you went with a camera only system the same percentage would find something to complain about. For now, I thinknodd of catching an issue as early as possible are better with hands on wheel. The onlynhesitation i have is how they do this in city noa where the wheel moves a lot more. Will have to test it (4 weeks, definitely ;) ).

When the system is more mature, or static, then a face sensor would be fine. In that way it is probably fine for Cadillac which has a very small, static, and well defined operational domain.