6 ms·

I appreciate Telsa's ambition to push technology forward but they shouldn't experiment with customers in real time like this. You want to replace parking sensor

I appreciate Telsa's ambition to push technology forward but they shouldn't experiment with customers in real time like this. You want to replace parking sensors with vision, go right ahead, prove it in the lab first, don't remove the parking sensors and just assume that the engineers will figure it out.

Eji1700 4y ago
Seeing this as "pushing technology forward" is a pretty generous interpretation. It reeks of cost cutting.
- MPSimmons 4y ago
  Agreed. I suspect there were parts shortages, and Elon was like, "The best part is no part, take it out of the design and do it with cameras". Same story with radar. Now both radar and sonar are coming back because it's far better to have the sensor data, even if you have to wade through it, than to not have it, particularly in adverse conditions.
  - gonehome 4y ago
    Karpathy gave a talk about radar - they were getting worse results with radar than without (according to him). Mostly because they would get false readings from radar shooting under the car and stuff like that - vision alone was more accurate.
    - ryanwaggoner 4y ago
      If you're talking about his appearance on Lex's podcast, he didn't say this. He said that the additional accuracy wasn't worth it. Which to me sounds like it was more accurate, but they didn't want to deal with the additional cost and complexity and thought that vision only would be good enough. But the link we're commenting on, along with the ongoing complaints about FSD, seem to indicate that the real world results haven't proven that theory out.
      - JoshCole 4y ago
        He is talking about the long technical talk at CVPR. https://www.youtube.com/watch?v=g6bOwQdCJrc https://www.youtube.com/watch?v=g6bOwQdCJrc For example look at this section: https://youtu.be/g6bOwQdCJrc?t=1370 https://youtu.be/g6bOwQdCJrc?t=1370 Also if you look into it you will find Google's self-driving car project ran into similar issues. They upgraded the radar in response to the issues rather than removing it, but there are actual issues. Vision only was better than radar plus vision in their stack. For real world situations that frequently come up in driving radar wasn't worth it. The theoretical accuracy of the sensor fusion didn't happen in practice because in practice you needed to have vision tell you when sensor fusion wasn't going to be accurate. It isn't naive sensor fusion as in toy math models of sensor fusion. So when there was a disagreement you needed to go with vision. Yet vision was already able to predict both the failure of and the actual result of radar. In theory, you could just have done the more complicated sensor fusion. In practice, the relative advantage of improving other parts of the vision stack far exceeded improving the sensor fusion. Over time one would expect the relative advantage to change as other improvements were made such that this was no longer true. That doesn't mean the time at the decision wasn't correct, it just means people who evaluate it with future changes already in place are suffering from hindsight bias/committing an anachronistic fallacy.
        gonehome 4y ago
        Yep - this is the one I was talking about, thanks for linking!
        ryanwaggoner 4y ago
        Yeah, Google actually made the right call. Thinking that Tesla made a stupid and dangerous call isn't "hindsight bias", their vision-only approach has always been inferior and reckless.
        JoshCole 4y ago
        The practical impact of the change you call stupid and reckless was that the car stopped slamming its brakes when going under underpasses. So when it comes down to it what you are claiming is that it is stupid and reckless to decide not to slam on the brakes. Look - I get it. I too thought radar sensor fusion would improve the results. Guess who else thought this? The people you consider stupid and reckless thought this. Do you know why they changed their mind? According to you, because they are stupid and reckless. I can't stress this enough: you are telling yourself inaccurate stories about what happened. If you just /look/ at the /measured impact/ instead of /wrongly guessing what Karpathy thought/ and /wrongly guessing what Karpathy has said/ and /wrongly guessing what other people were referencing when they talked about what Karpathy said/ then you would realize this.
    - pclmulqdq 4y ago
      Later, Musk came out and said that they were using cheap radar sensors, and that higher-quality radar sensors would likely have been better than vision alone.
      - judge2020 4y ago
        But radar only does anything for detecting cars in front of you, so the only deficiency by removing it is using vision to perform that distance-keeping when using traffic-aware cruise control, which is not a pain point in my 2022 M3LR even in the rain.
        saurik 4y ago
        But it used to be better: it was able to follow closer to the car in front--which is really important when you are in stop-and-go 3mph traffic on the highway--and it was supposedly able to see under the car in front of you (reflecting off the highway) to track the car ahead of them <- a friend of mine who's entire family is super into Tesla has all cars with the radar sensors, and when I was driving her around she was commenting about how many dumb decisions the newer Model 3 I have been renting was making, seemingly because of this. She was not happy about it when she heard Tesla started deactivating radar in old cars. But yeah: the feature minimally seems to work well enough--...even surprisingly well!--given the lack of having the sensors that Tesla previously swore were amazing, and so in some sense that particular cost cutting measure was a success. I was even thinking of buying one to own... but then they removed the parking sensors, and frankly that feature only barely works WITH the sensors--I drive a ton of random cars as I am always renting stuff, and the Tesla's parking radius feature is a cool UX but has tons of slop and has massively high latency... it is frankly just super dumb--and I cancelled my order when it finally arrived a week after it was for sure I wouldn't have the sensors. Regardless, though, even if you feel it works perfectly, radar was another situation of just hastily scrapping stuff and working on it in the field, as they removed the sensor before they had the software which was tolerable, and then they just kept promising people it was going to get better as they frantically tried to turn their marketing into reality. Over time it got better and after a long time it finally started to work at higher speeds, but for a long time it had a cap lower than the in-practice speed of even my nearby highways here in the United States, and still hasn't managed to regain all of the functionality of the prior design. Can you imagine Apple removing a sensor in a new iPhone without already having the software to replace it, and then just promising new customers they'll figure it out in the hope it doesn't affect sales?
  - ModernMech 4y ago
    > Now both radar and sonar are coming back because it's far better to have the sensor data, even if you have to wade through it, than to not have it, particularly in adverse conditions. This is something that every roboticist has known for 60+ years. It's just galling to me that two people had to lose their heads because new born Tesla Inc. and the know-nothing manchild running it, felt they had the right (due to the size of their wallets, not their knowledge or experience) to overrule people who knew better. And people call this "pushing technology forward".
- Someone1234 4y ago
  Agreed. Actually pushing technology forward would involve replacing parking sensors with two additional cameras to cover the blind-spot, then using vision to exceed what parking sensors could do. It isn't simply to remove parking sensors, have areas of the vehicle with blind spots, then hack the problem by trying to "guess" what in happening in those locations. It has been suggested this removal saved Tesla less than $150/car. That's crazy for a $40K+ vehicle. There's nothing to defend here, it makes their vehicles objectively worse.
  - saurik 4y ago
    FWIW, while I am in the camp if people who think this was a dumb move and who in fact cancelled my buying a Tesla over it at the last possible moment I could still walk away from the car, I had thought Tesla wasn't just running into cost issues: they also were also quickly reacting to the semiconductor shortage and even had tons of cars sitting around missing just a few parts as they frantically tried to ship, and so maybe dropping this part wasn't as much "cutting costs" as "cutting dependencies".
  - inferiorhuman 4y ago
    It has been suggested this removal saved Tesla less than $150/car. Far be it for me to defend lord elno, but $150 is a pretty hefty amount in automotive circles. Car manufacturers (especially the Big 3) do all sorts of shit to save literal pennies on unit cost. The idea is that their cars are sold in such high volume they're saving a ton of money. Of course it's myopic since that doesn't account for customers' willingness to pay a bit more here and there for a higher quality product.
- jeffbee 4y ago
  Tesla is way behind everyone on computer vision. They have not "pushed the technology forward" at any time in any respect.
  - amluto 4y ago
    Having tested a modern Tesla car a few weeks ago, I was mostly impressed by how bad the vision algorithm was at understanding the roads. There were plenty of entirely fixed curbs, intersections, lane lines, etc that it could not handle correctly, and this was on a clear, dry day with excellent visibility. They’ve certainly pushed the willingness to use so-so technology forward, but I was really not impressed by the underlying technology. The odd thing is that these type of errors really ought to be correctable by training with a small fleet of humans annotating data. AI is very, very good at getting examples from the training set right, and fixed roads a few tens of miles from the old headquarters ought to be in or at least very similar to the training data.
- kiratp 4y ago
  Cost cutting is a good thing if it makes things cheaper and more accessible. Yes, I know that in the short term Tesla will see expanded margins but longer term this will make things cheaper.
  - i_am_jl 4y ago
    All other things equal, sure, but this isn't just cheaper and more accessible; it's cheaper and more accessible and worse.
  - sosodev 4y ago
    Most of the savings from cost cuts go back to the company. The buyer in this case gets a highly degraded experience with negligible savings.
- moogly 4y ago
  I don't really see how putting a bunch of Californian developers on the task to write software to solve a problem that $114/car (according to an estimate) already solved could be considered cutting costs. They've already spent 6 months on 1 out of 5 features they lost by getting rid of their sensors.
  - rootusrootus 4y ago
    This is the same company that continues to waste countless hours of development time trying to get a camera-based solution to an IR rain sensor that costs about 5 bucks. I think there is a strong ideological component, not just cost cutting.
    - pclmulqdq 4y ago
      The biggest Wall Street "bull case" for Tesla these days is around the gross margin on their cars. Tesla has always been a little bit of an accounting project (around subsidies, etc.), and this sounds like the kind of thing that a company with a keen awareness of their stock price and a lot of comfort with accounting sleight-of-hand would do. The accountants capitalize a huge R&D loss, which fits with the company being "innovative," and in return the gross margin of the car is higher.
  - tpmx 4y ago
    Especially when the software-based solution very likely will fail in weather that is very unusual in California but relatively common elsewhere.
onlyrealcuzzo 4y ago
How much do these sensors even cost? My understanding is not much. It's amazing what Tesla can get away with and keep this reputation of being the best company to ever grace the surface of the Earth.
- blibble 4y ago
  about $2 each
SQueeeeeL 4y ago
They have blatantly shown a pretty large disregard for consumer safety multiple times, this really isn't surprising. Not using lidar and releasing "self-driving" modes which routinely caused accidents both come to mind. Telsa is basically done now that real players have entered the EV market, you can kinda see it with the upper managements strange sudden shift to social media operation
- jackmott42 4y ago
  > Telsa is basically done now Its been a long time since the other players entered the market and they still have a horrible state with their charging networks. There are many non Teslas I would like to buy but none are viable options unless its a car that will stay in town only. I know electrify America is "ok", I've used it, multiple places with success, though always at least some stalls are broken, in weird ways, and its less pleasant to use even if it works.
  - gambiting 4y ago
    Fortunately US isn't the only country in the world - elsewhere Tesla is failing to take hold, it's just way too expensive for what it is. No idea why you'd buy a Model 3 over a Skoda Enyaq for instance.
    - brandonagr2 4y ago
      Are you sure about that? https://www.torquenews.com/15475/europe-embraces-tesla-model-y-record-breaking-sales-february-2023 https://www.torquenews.com/15475/europe-embraces-tesla-model... "if this trend continues, the Model Y could easily secure a place in the top 10 best-selling cars in Europe by the end of the year"
- brandonagr2 4y ago
  Where have you see FSD routinely causing accidents?
jandrese 4y ago
Full Self Driving has arguably been regressing as they try to do everything with just a few cameras as well. Here's what self proclaimed Tesla fanboys discovered when they tried the newest FSD beta: https://www.youtube.com/watch?v=9ND2ibwue_E https://www.youtube.com/watch?v=9ND2ibwue_E You will often see people saying that they've put thousands of miles on their Tesla without incident, but pretty much inevitably when you dig into it you will discover that those were all highway miles where it's little more than spiffy cruise control with basic lanekeeping.
- varispeed 4y ago
  There is inescapable fact that the AI at its current state can't reason and it can only deal with problems it had been trained on. Which means that on the road, where _anything_ can happen, there is a ton of blind spots where the AI rather than taking a reasonable decision, would hallucinate one, that could turn out to be disastrous. The current solution is to train for every possible scenario and hope for the best, but I don't think that should cut it when the human lives are at stake.
  - iknowstuff 4y ago
    You’re talking out your butt. That’s not how these cars make decisions. It’s not AI in the deep/learning neural nets sense making decisions.
    - varispeed 4y ago
      Such an eloquent reply, well done. ChatGPT: Tesla's Autopilot system uses a variety of machine learning algorithms to enable its autonomous driving capabilities. Here are a few examples: Convolutional Neural Networks (CNNs): These are commonly used in computer vision applications and are particularly useful for object recognition. In the context of Tesla's Autopilot system, CNNs are used to identify and track objects on the road, such as other vehicles, pedestrians, and road signs. Recurrent Neural Networks (RNNs): These are commonly used in natural language processing and sequential data analysis. In the context of Tesla's Autopilot system, RNNs are used to predict the behavior of other vehicles on the road and to make decisions about how the Tesla vehicle should respond. Decision Trees: These are commonly used in classification and regression tasks. In the context of Tesla's Autopilot system, decision trees are used to make decisions about the behavior of the Tesla vehicle based on its sensors and other inputs. Reinforcement Learning: This is a type of machine learning that involves an agent (in this case, the Tesla vehicle) learning to make decisions based on feedback from its environment. In the context of Tesla's Autopilot system, reinforcement learning is used to teach the car how to navigate complex driving scenarios, such as merging onto a busy highway or navigating a crowded intersection. Overall, Tesla's Autopilot system uses a combination of these and other machine learning algorithms to enable its autonomous driving capabilities, with the goal of providing a safer and more efficient driving experience.
jackmott42 4y ago
You still can't use cruise control on a tesla safely because they refuse to put in a normal cruise control option and their traffic aware one still slams the brakes inappropriate at intermittent times. And careful! If you are about to tell me YOUR Tesla cruise control works fine, it is just a matter of time before it doesn't. We have all been there. auto wipers are still also wonky because the ambition is to do it with vision but they haven't figured that out yet either.
- breck 4y ago
  I love the Autosteer. I just did a cross country trip with it. I did get 2 or 3 of those brake slamming moments though (they keep it...interesting?). Auto wipers aren't there yet. I agree with you. Also, often when I would say "Wipers off", it would try to navigate to "Microsoft". The voice recognition is not quite ChatGPT smart.
  - rootusrootus 4y ago
    I rest my foot on the accelerator pedal so that I can react quickly when using AP. "Interesting" translates to "my wife gets irritated when I engage AP" because the occasional phantom braking destroys any sort of confidence she has in the car. > Auto wipers aren't there yet. They are better than my first Model 3 in 2019. But not that much better, and it is 2023. Auto wipers will work on a Tesla as soon as they start using the ubiquitous $5 IR sensor everyone else has been using for the last few decades, and not a moment sooner. > The voice recognition is Incredibly shitty, that's what it is. Let's just admit that. It gets used as an excuse for all the missing controls, but it's pretty awful. They should just offer a retrofit stalk that has proper wiper controls on it. On most other mainstream manufacturers that would be easy, the stalk is just a module that hooks into the CAN bus and is replaceable with two screws.
  - chankstein38 4y ago
    ChatGPT is not a speech dictation engine.
  - robbiep 4y ago
    > I did get 2 or 3 of those brake slamming moments though (they keep it...interesting?). Not a word you want to describe your car trip?
kypro 4y ago
> You want to replace parking sensors with vision, go right ahead, prove it in the lab first, don't remove the parking sensors and just assume that the engineers will figure it out. That's what surprised me about this. Some of these tests were really obvious. Failing one or two edge cases is forgivable, but being this inaccurate so consistently says far more about the company and their priorities. Clearly they are not QAing their products properly. This is a customer facing feature so it's easy for a reviewing to identify and test, but they're willing to cut costs on a safety feature like this then who knows what else they're doing to save costs that isn't so obvious.

Eji1700 4y ago

Seeing this as "pushing technology forward" is a pretty generous interpretation. It reeks of cost cutting.

MPSimmons 4y ago
Agreed. I suspect there were parts shortages, and Elon was like, "The best part is no part, take it out of the design and do it with cameras". Same story with radar. Now both radar and sonar are coming back because it's far better to have the sensor data, even if you have to wade through it, than to not have it, particularly in adverse conditions.
- gonehome 4y ago
  Karpathy gave a talk about radar - they were getting worse results with radar than without (according to him). Mostly because they would get false readings from radar shooting under the car and stuff like that - vision alone was more accurate.
  - ryanwaggoner 4y ago
    If you're talking about his appearance on Lex's podcast, he didn't say this. He said that the additional accuracy wasn't worth it. Which to me sounds like it was more accurate, but they didn't want to deal with the additional cost and complexity and thought that vision only would be good enough. But the link we're commenting on, along with the ongoing complaints about FSD, seem to indicate that the real world results haven't proven that theory out.
    - JoshCole 4y ago
      He is talking about the long technical talk at CVPR. https://www.youtube.com/watch?v=g6bOwQdCJrc https://www.youtube.com/watch?v=g6bOwQdCJrc For example look at this section: https://youtu.be/g6bOwQdCJrc?t=1370 https://youtu.be/g6bOwQdCJrc?t=1370 Also if you look into it you will find Google's self-driving car project ran into similar issues. They upgraded the radar in response to the issues rather than removing it, but there are actual issues. Vision only was better than radar plus vision in their stack. For real world situations that frequently come up in driving radar wasn't worth it. The theoretical accuracy of the sensor fusion didn't happen in practice because in practice you needed to have vision tell you when sensor fusion wasn't going to be accurate. It isn't naive sensor fusion as in toy math models of sensor fusion. So when there was a disagreement you needed to go with vision. Yet vision was already able to predict both the failure of and the actual result of radar. In theory, you could just have done the more complicated sensor fusion. In practice, the relative advantage of improving other parts of the vision stack far exceeded improving the sensor fusion. Over time one would expect the relative advantage to change as other improvements were made such that this was no longer true. That doesn't mean the time at the decision wasn't correct, it just means people who evaluate it with future changes already in place are suffering from hindsight bias/committing an anachronistic fallacy.
      - gonehome 4y ago
        Yep - this is the one I was talking about, thanks for linking!
      - ryanwaggoner 4y ago
        Yeah, Google actually made the right call. Thinking that Tesla made a stupid and dangerous call isn't "hindsight bias", their vision-only approach has always been inferior and reckless.
        JoshCole 4y ago
        The practical impact of the change you call stupid and reckless was that the car stopped slamming its brakes when going under underpasses. So when it comes down to it what you are claiming is that it is stupid and reckless to decide not to slam on the brakes. Look - I get it. I too thought radar sensor fusion would improve the results. Guess who else thought this? The people you consider stupid and reckless thought this. Do you know why they changed their mind? According to you, because they are stupid and reckless. I can't stress this enough: you are telling yourself inaccurate stories about what happened. If you just /look/ at the /measured impact/ instead of /wrongly guessing what Karpathy thought/ and /wrongly guessing what Karpathy has said/ and /wrongly guessing what other people were referencing when they talked about what Karpathy said/ then you would realize this.
  - pclmulqdq 4y ago
    Later, Musk came out and said that they were using cheap radar sensors, and that higher-quality radar sensors would likely have been better than vision alone.
    - judge2020 4y ago
      But radar only does anything for detecting cars in front of you, so the only deficiency by removing it is using vision to perform that distance-keeping when using traffic-aware cruise control, which is not a pain point in my 2022 M3LR even in the rain.
      - saurik 4y ago
        But it used to be better: it was able to follow closer to the car in front--which is really important when you are in stop-and-go 3mph traffic on the highway--and it was supposedly able to see under the car in front of you (reflecting off the highway) to track the car ahead of them <- a friend of mine who's entire family is super into Tesla has all cars with the radar sensors, and when I was driving her around she was commenting about how many dumb decisions the newer Model 3 I have been renting was making, seemingly because of this. She was not happy about it when she heard Tesla started deactivating radar in old cars. But yeah: the feature minimally seems to work well enough--...even surprisingly well!--given the lack of having the sensors that Tesla previously swore were amazing, and so in some sense that particular cost cutting measure was a success. I was even thinking of buying one to own... but then they removed the parking sensors, and frankly that feature only barely works WITH the sensors--I drive a ton of random cars as I am always renting stuff, and the Tesla's parking radius feature is a cool UX but has tons of slop and has massively high latency... it is frankly just super dumb--and I cancelled my order when it finally arrived a week after it was for sure I wouldn't have the sensors. Regardless, though, even if you feel it works perfectly, radar was another situation of just hastily scrapping stuff and working on it in the field, as they removed the sensor before they had the software which was tolerable, and then they just kept promising people it was going to get better as they frantically tried to turn their marketing into reality. Over time it got better and after a long time it finally started to work at higher speeds, but for a long time it had a cap lower than the in-practice speed of even my nearby highways here in the United States, and still hasn't managed to regain all of the functionality of the prior design. Can you imagine Apple removing a sensor in a new iPhone without already having the software to replace it, and then just promising new customers they'll figure it out in the hope it doesn't affect sales?
- ModernMech 4y ago
  > Now both radar and sonar are coming back because it's far better to have the sensor data, even if you have to wade through it, than to not have it, particularly in adverse conditions. This is something that every roboticist has known for 60+ years. It's just galling to me that two people had to lose their heads because new born Tesla Inc. and the know-nothing manchild running it, felt they had the right (due to the size of their wallets, not their knowledge or experience) to overrule people who knew better. And people call this "pushing technology forward".
Someone1234 4y ago
Agreed. Actually pushing technology forward would involve replacing parking sensors with two additional cameras to cover the blind-spot, then using vision to exceed what parking sensors could do. It isn't simply to remove parking sensors, have areas of the vehicle with blind spots, then hack the problem by trying to "guess" what in happening in those locations. It has been suggested this removal saved Tesla less than $150/car. That's crazy for a $40K+ vehicle. There's nothing to defend here, it makes their vehicles objectively worse.
- saurik 4y ago
  FWIW, while I am in the camp if people who think this was a dumb move and who in fact cancelled my buying a Tesla over it at the last possible moment I could still walk away from the car, I had thought Tesla wasn't just running into cost issues: they also were also quickly reacting to the semiconductor shortage and even had tons of cars sitting around missing just a few parts as they frantically tried to ship, and so maybe dropping this part wasn't as much "cutting costs" as "cutting dependencies".
- inferiorhuman 4y ago
  It has been suggested this removal saved Tesla less than $150/car. Far be it for me to defend lord elno, but $150 is a pretty hefty amount in automotive circles. Car manufacturers (especially the Big 3) do all sorts of shit to save literal pennies on unit cost. The idea is that their cars are sold in such high volume they're saving a ton of money. Of course it's myopic since that doesn't account for customers' willingness to pay a bit more here and there for a higher quality product.
jeffbee 4y ago
Tesla is way behind everyone on computer vision. They have not "pushed the technology forward" at any time in any respect.
- amluto 4y ago
  Having tested a modern Tesla car a few weeks ago, I was mostly impressed by how bad the vision algorithm was at understanding the roads. There were plenty of entirely fixed curbs, intersections, lane lines, etc that it could not handle correctly, and this was on a clear, dry day with excellent visibility. They’ve certainly pushed the willingness to use so-so technology forward, but I was really not impressed by the underlying technology. The odd thing is that these type of errors really ought to be correctable by training with a small fleet of humans annotating data. AI is very, very good at getting examples from the training set right, and fixed roads a few tens of miles from the old headquarters ought to be in or at least very similar to the training data.
kiratp 4y ago
Cost cutting is a good thing if it makes things cheaper and more accessible. Yes, I know that in the short term Tesla will see expanded margins but longer term this will make things cheaper.
- i_am_jl 4y ago
  All other things equal, sure, but this isn't just cheaper and more accessible; it's cheaper and more accessible and worse.
- sosodev 4y ago
  Most of the savings from cost cuts go back to the company. The buyer in this case gets a highly degraded experience with negligible savings.
moogly 4y ago
I don't really see how putting a bunch of Californian developers on the task to write software to solve a problem that $114/car (according to an estimate) already solved could be considered cutting costs. They've already spent 6 months on 1 out of 5 features they lost by getting rid of their sensors.
- rootusrootus 4y ago
  This is the same company that continues to waste countless hours of development time trying to get a camera-based solution to an IR rain sensor that costs about 5 bucks. I think there is a strong ideological component, not just cost cutting.
  - pclmulqdq 4y ago
    The biggest Wall Street "bull case" for Tesla these days is around the gross margin on their cars. Tesla has always been a little bit of an accounting project (around subsidies, etc.), and this sounds like the kind of thing that a company with a keen awareness of their stock price and a lot of comfort with accounting sleight-of-hand would do. The accountants capitalize a huge R&D loss, which fits with the company being "innovative," and in return the gross margin of the car is higher.
- tpmx 4y ago
  Especially when the software-based solution very likely will fail in weather that is very unusual in California but relatively common elsewhere.

onlyrealcuzzo 4y ago

How much do these sensors even cost? My understanding is not much. It's amazing what Tesla can get away with and keep this reputation of being the best company to ever grace the surface of the Earth.

blibble 4y ago
about $2 each

SQueeeeeL 4y ago

They have blatantly shown a pretty large disregard for consumer safety multiple times, this really isn't surprising. Not using lidar and releasing "self-driving" modes which routinely caused accidents both come to mind. Telsa is basically done now that real players have entered the EV market, you can kinda see it with the upper managements strange sudden shift to social media operation

jackmott42 4y ago
> Telsa is basically done now Its been a long time since the other players entered the market and they still have a horrible state with their charging networks. There are many non Teslas I would like to buy but none are viable options unless its a car that will stay in town only. I know electrify America is "ok", I've used it, multiple places with success, though always at least some stalls are broken, in weird ways, and its less pleasant to use even if it works.
- gambiting 4y ago
  Fortunately US isn't the only country in the world - elsewhere Tesla is failing to take hold, it's just way too expensive for what it is. No idea why you'd buy a Model 3 over a Skoda Enyaq for instance.
  - brandonagr2 4y ago
    Are you sure about that? https://www.torquenews.com/15475/europe-embraces-tesla-model-y-record-breaking-sales-february-2023 https://www.torquenews.com/15475/europe-embraces-tesla-model... "if this trend continues, the Model Y could easily secure a place in the top 10 best-selling cars in Europe by the end of the year"
brandonagr2 4y ago
Where have you see FSD routinely causing accidents?

jandrese 4y ago

Full Self Driving has arguably been regressing as they try to do everything with just a few cameras as well. Here's what self proclaimed Tesla fanboys discovered when they tried the newest FSD beta: https://www.youtube.com/watch?v=9ND2ibwue_E https://www.youtube.com/watch?v=9ND2ibwue_E You will often see people saying that they've put thousands of miles on their Tesla without incident, but pretty much inevitably when you dig into it you will discover that those were all highway miles where it's little more than spiffy cruise control with basic lanekeeping.

varispeed 4y ago
There is inescapable fact that the AI at its current state can't reason and it can only deal with problems it had been trained on. Which means that on the road, where _anything_ can happen, there is a ton of blind spots where the AI rather than taking a reasonable decision, would hallucinate one, that could turn out to be disastrous. The current solution is to train for every possible scenario and hope for the best, but I don't think that should cut it when the human lives are at stake.
- iknowstuff 4y ago
  You’re talking out your butt. That’s not how these cars make decisions. It’s not AI in the deep/learning neural nets sense making decisions.
  - varispeed 4y ago
    Such an eloquent reply, well done. ChatGPT: Tesla's Autopilot system uses a variety of machine learning algorithms to enable its autonomous driving capabilities. Here are a few examples: Convolutional Neural Networks (CNNs): These are commonly used in computer vision applications and are particularly useful for object recognition. In the context of Tesla's Autopilot system, CNNs are used to identify and track objects on the road, such as other vehicles, pedestrians, and road signs. Recurrent Neural Networks (RNNs): These are commonly used in natural language processing and sequential data analysis. In the context of Tesla's Autopilot system, RNNs are used to predict the behavior of other vehicles on the road and to make decisions about how the Tesla vehicle should respond. Decision Trees: These are commonly used in classification and regression tasks. In the context of Tesla's Autopilot system, decision trees are used to make decisions about the behavior of the Tesla vehicle based on its sensors and other inputs. Reinforcement Learning: This is a type of machine learning that involves an agent (in this case, the Tesla vehicle) learning to make decisions based on feedback from its environment. In the context of Tesla's Autopilot system, reinforcement learning is used to teach the car how to navigate complex driving scenarios, such as merging onto a busy highway or navigating a crowded intersection. Overall, Tesla's Autopilot system uses a combination of these and other machine learning algorithms to enable its autonomous driving capabilities, with the goal of providing a safer and more efficient driving experience.

jackmott42 4y ago

You still can't use cruise control on a tesla safely because they refuse to put in a normal cruise control option and their traffic aware one still slams the brakes inappropriate at intermittent times. And careful! If you are about to tell me YOUR Tesla cruise control works fine, it is just a matter of time before it doesn't. We have all been there. auto wipers are still also wonky because the ambition is to do it with vision but they haven't figured that out yet either.

breck 4y ago
I love the Autosteer. I just did a cross country trip with it. I did get 2 or 3 of those brake slamming moments though (they keep it...interesting?). Auto wipers aren't there yet. I agree with you. Also, often when I would say "Wipers off", it would try to navigate to "Microsoft". The voice recognition is not quite ChatGPT smart.
- rootusrootus 4y ago
  I rest my foot on the accelerator pedal so that I can react quickly when using AP. "Interesting" translates to "my wife gets irritated when I engage AP" because the occasional phantom braking destroys any sort of confidence she has in the car. > Auto wipers aren't there yet. They are better than my first Model 3 in 2019. But not that much better, and it is 2023. Auto wipers will work on a Tesla as soon as they start using the ubiquitous $5 IR sensor everyone else has been using for the last few decades, and not a moment sooner. > The voice recognition is Incredibly shitty, that's what it is. Let's just admit that. It gets used as an excuse for all the missing controls, but it's pretty awful. They should just offer a retrofit stalk that has proper wiper controls on it. On most other mainstream manufacturers that would be easy, the stalk is just a module that hooks into the CAN bus and is replaceable with two screws.
- chankstein38 4y ago
  ChatGPT is not a speech dictation engine.
- robbiep 4y ago
  > I did get 2 or 3 of those brake slamming moments though (they keep it...interesting?). Not a word you want to describe your car trip?

kypro 4y ago

> You want to replace parking sensors with vision, go right ahead, prove it in the lab first, don't remove the parking sensors and just assume that the engineers will figure it out. That's what surprised me about this. Some of these tests were really obvious. Failing one or two edge cases is forgivable, but being this inaccurate so consistently says far more about the company and their priorities. Clearly they are not QAing their products properly. This is a customer facing feature so it's easy for a reviewing to identify and test, but they're willing to cut costs on a safety feature like this then who knows what else they're doing to save costs that isn't so obvious.