5 ms·

Tesla's system is generalised and can operate on any roads. It also uses vision, as opposed to a crazy number of sensors including LIDAR. Tesla FSD is a much mo

by cbeach 3y ago

Tesla's system is generalised and can operate on any roads. It also uses vision, as opposed to a crazy number of sensors including LIDAR. Tesla FSD is a much more promising and human-like AI.

aeyes 3y ago
If, and that is a big IF, they manage to achieve it. My best guess is that Tesla won't get anywhere close to L5 before the end of this decade. Your best chance today is to pack more sensores. Edit: I still believe that the systems which assist the driver today are useful and can make driving safer and easier. I don't want to downplay what they have achieved but they are a looong way from L5.
- tsimionescu 3y ago
  I believe the CEO of Waymo, the company that is actually succeeding in self-driving, has said several times now that L5 is not even on the horizon, for anyone. The problem isn't just training, it's sensors. There are simply no sensors available today that work in all the different weather conditions that L5 requires with the required precision.
- mythhabit 3y ago
  And even if they manage to pull it off, they ought to have data showing that they, in particular, can remove lidar sensor without losing fidelity, simply because we know that lidar is what gives reliable range information.
blitz_skull 3y ago
True FSD will only come from LIDAR + vision. Relying only on vision feels exceedingly dangerous. I recall a recent (last 2 years?) case where a guy was essentially decapitated by a flat-bed semi because his car couldn’t recognize the semi’s bed due to it being right at camera level. I think the Pure vision stuff is nifty, but “nifty” is the exact dead-last parameter when trying to transport my family safely from point A to point B.
- appplication 3y ago
  Vision will never be the solution, autonomous vehicles need ranging capability, and vision does not allow that in a foolproof capacity.
  - cbeach 3y ago
    So how does LIDAR read a sign (not just the shape of the sign, but the content)? How does it read the colour of a traffic light? How does it read brake light status or indicators on the car in front? Oh, it can't? So you need cameras too? So you have to build a 3-dimensional sensor fusion system for LIDAR to work? Wouldn't that fusion system be more complex, less performant, and more fallible that just choosing a single model (vision/LIDAR) and optimising around that? 10 million years of animal evolution makes a good case for stereoscopic vision as a sensing system for navigating the world. LIDAR is a crutch for low-maturity software systems.
    - microtherion 3y ago
      > 10 million years of animal evolution makes a good case for stereoscopic vision Stereoscopic vision? That's when you apply sensor fusion to two cameras (and a number of microphones, too), isn't it? If airplane designers had trusted in "animal evolution", planes would flap their wings to gain height.
    - jolux 3y ago
      Somehow I feel Tesla stock would not cost what it does if Musk said FSD was only 10 million years away, just be patient!
    - tsimionescu 3y ago
      Not all animals have stereoscopic vision. Many birds and fish see entirely separate images from their two eyes. Animals with injuries to one eye still have functional vision, even if worse. The "brain" part is the one that really helps with animal vision. And it includes evolved generational models, sensor fusion, memories of past experiences and other inputs beyond just two "cameras". Overall, the complexity of modeling the world the way an animal does seems much much bigger than a few different sensors.
ben_w 3y ago
Are "human-like" and vision-only supposed to be good things? I don't care how the magic box does what it does, only how well it does it.
wkat4242 3y ago
Humans are pretty fallible though. And human eyes are way better in dynamic lighting conditions than cameras. Our dynamic range is huge in comparison. Elon's objections to lidar are not because it's a bad idea but because of the costs. Especially because he's already sold cars with "full self driving" that don't have LiDAR. So he's painted himself in a corner. Even if it can be done with vision only, lidar will add an extra data point for cases where the object recognition is in doubt. And the price will come down.
- actionfromafar 3y ago
  Humans have great sensor fusion, too, with stereo microphones in most models, as well as situational awareness. One should think of the car cameras as a fixed periscope view. Try driving with that, what the computer sees is more like driving a tank with monitors. Not the best. It’s only reasonable to give the cars some super human senses to compensate, like LIDAR etc.
- xorcist 3y ago
  It like when Apple funded lots of usability studies on why one mouse button was optimal for most users. They held out longer than expected, some 20 years or so, but common sense prevailed. Even if they had to make a design where the buttons weren't clearly visible first.
  - wkat4242 3y ago
    Yeah I think part of that was Steve Jobs' extreme aversion to being wrong. And he of course shared a lot of personality traits with Musk (narcissism in particular) so that makes sense. Even today Macs don't really have 2 mouse buttons, they just call a double-finger click a "gesture" even though it is simply a right-click.
actionfromafar 3y ago
So far, broken promises.
flohofwoe 3y ago
> Tesla FSD is a much more promising and human-like AI. This can be true without both ever materializing though ;)
cmsonger 3y ago
I'm not sure: "I drove with the radio blaring so I could not hear the siren" is something to brag about. More sensors == better. And the fact that LiDAR costs have come down two orders of magnitude since Tesla made this decision suggests that some of the economic drivers for such a decision have changed. Part of the reason I've owned FSD since 2017 and yet spent exactly zero minutes in my car in the expressway, in light traffic, in good weather, in the middle of the day reading a book is because of the lack of an entry on the posted timeline saying something like: "Tesla states it will accept liability for any crashes that occur while FSD was engaged." After all these years, it's fair to say: "In the 7 years that Tesla has accepted money from customers for FSD, Tesla still requires human supervision in all cases." So far they have proven they are asmpytotically approaching something that drives badly and is in danger of hitting things.
abenga 3y ago
Why is more sensors less promising?
trabant00 3y ago
My system is even better, it requires no additional hardware whatsoever, nor any single line of code. And it's not only for cars but any vehicle in any environment. It doesn't work for now, but that only means it's in the same state as the Tesla system, meaning very promising.
hnbad 3y ago
This must be a troll comment. Humans don't rely on vision when driving because it's superior, humans rely on vision because that's all we have. Actually that isn't even true as we also use our ears and with assisted driving also the car's various sensors (including LIDAR). The only justification for sticking to vision alone is nerd cred. We don't want human-like AI, we want safe AI. Right now Tesla's AI is making mistakes no human would ever make so it's not even getting that part right. Not using a "crazy number of sensors" is not something you can brag about outside a kind of "sell me this pen" sales pitch.
klabb3 3y ago
I don’t know the specifics and have no dog in the fight, but the argument is flawed here. Human eye sight is in some ways better than small form factor cameras, eg anti-glaring, dynamic range, light sensitivity at night. Also, our vision is quite bad at some stuff, for instance distance estimates have a high error rate. You can do a lot better than that. Tech is not biology, so in order for a computer to not do something stupid, it generally needs different sensors. Some sensors (depth gauge when scuba diving, altimeters for planes, or even gps) are obvious examples of when tech outperforms humans and is safety critical. Just because humans can drive doesn’t mean computers should replicate our “control loop”.
- tsimionescu 3y ago
  > Also, our vision is quite bad at some stuff, for instance distance estimates have a high error rate. You can do a lot better than that. Our distance estimates are quite good at human scales, especially when compared to vision based approaches. We don't have great precision (we will never say something like "it's 19.7m away") but we very rarely grossly misestimate (think something is 10m away when it's actually 200m away or 10cm away). Of course, compared to a LIDAR or advanced techniques used in physics, we don't hold the tiniest candles (say, if we were to compare to LIGO's distance measurements).

by cbeach 3y ago

Tesla's system is generalised and can operate on any roads. It also uses vision, as opposed to a crazy number of sensors including LIDAR. Tesla FSD is a much more promising and human-like AI.

aeyes 3y ago

If, and that is a big IF, they manage to achieve it. My best guess is that Tesla won't get anywhere close to L5 before the end of this decade. Your best chance today is to pack more sensores. Edit: I still believe that the systems which assist the driver today are useful and can make driving safer and easier. I don't want to downplay what they have achieved but they are a looong way from L5.

tsimionescu 3y ago
I believe the CEO of Waymo, the company that is actually succeeding in self-driving, has said several times now that L5 is not even on the horizon, for anyone. The problem isn't just training, it's sensors. There are simply no sensors available today that work in all the different weather conditions that L5 requires with the required precision.
mythhabit 3y ago
And even if they manage to pull it off, they ought to have data showing that they, in particular, can remove lidar sensor without losing fidelity, simply because we know that lidar is what gives reliable range information.

blitz_skull 3y ago

True FSD will only come from LIDAR + vision. Relying only on vision feels exceedingly dangerous. I recall a recent (last 2 years?) case where a guy was essentially decapitated by a flat-bed semi because his car couldn’t recognize the semi’s bed due to it being right at camera level. I think the Pure vision stuff is nifty, but “nifty” is the exact dead-last parameter when trying to transport my family safely from point A to point B.

appplication 3y ago
Vision will never be the solution, autonomous vehicles need ranging capability, and vision does not allow that in a foolproof capacity.
- cbeach 3y ago
  So how does LIDAR read a sign (not just the shape of the sign, but the content)? How does it read the colour of a traffic light? How does it read brake light status or indicators on the car in front? Oh, it can't? So you need cameras too? So you have to build a 3-dimensional sensor fusion system for LIDAR to work? Wouldn't that fusion system be more complex, less performant, and more fallible that just choosing a single model (vision/LIDAR) and optimising around that? 10 million years of animal evolution makes a good case for stereoscopic vision as a sensing system for navigating the world. LIDAR is a crutch for low-maturity software systems.
  - microtherion 3y ago
    > 10 million years of animal evolution makes a good case for stereoscopic vision Stereoscopic vision? That's when you apply sensor fusion to two cameras (and a number of microphones, too), isn't it? If airplane designers had trusted in "animal evolution", planes would flap their wings to gain height.
  - jolux 3y ago
    Somehow I feel Tesla stock would not cost what it does if Musk said FSD was only 10 million years away, just be patient!
  - tsimionescu 3y ago
    Not all animals have stereoscopic vision. Many birds and fish see entirely separate images from their two eyes. Animals with injuries to one eye still have functional vision, even if worse. The "brain" part is the one that really helps with animal vision. And it includes evolved generational models, sensor fusion, memories of past experiences and other inputs beyond just two "cameras". Overall, the complexity of modeling the world the way an animal does seems much much bigger than a few different sensors.

ben_w 3y ago

Are "human-like" and vision-only supposed to be good things? I don't care how the magic box does what it does, only how well it does it.

wkat4242 3y ago

Humans are pretty fallible though. And human eyes are way better in dynamic lighting conditions than cameras. Our dynamic range is huge in comparison. Elon's objections to lidar are not because it's a bad idea but because of the costs. Especially because he's already sold cars with "full self driving" that don't have LiDAR. So he's painted himself in a corner. Even if it can be done with vision only, lidar will add an extra data point for cases where the object recognition is in doubt. And the price will come down.

actionfromafar 3y ago
Humans have great sensor fusion, too, with stereo microphones in most models, as well as situational awareness. One should think of the car cameras as a fixed periscope view. Try driving with that, what the computer sees is more like driving a tank with monitors. Not the best. It’s only reasonable to give the cars some super human senses to compensate, like LIDAR etc.
xorcist 3y ago
It like when Apple funded lots of usability studies on why one mouse button was optimal for most users. They held out longer than expected, some 20 years or so, but common sense prevailed. Even if they had to make a design where the buttons weren't clearly visible first.
- wkat4242 3y ago
  Yeah I think part of that was Steve Jobs' extreme aversion to being wrong. And he of course shared a lot of personality traits with Musk (narcissism in particular) so that makes sense. Even today Macs don't really have 2 mouse buttons, they just call a double-finger click a "gesture" even though it is simply a right-click.

actionfromafar 3y ago

So far, broken promises.

flohofwoe 3y ago

> Tesla FSD is a much more promising and human-like AI. This can be true without both ever materializing though ;)

cmsonger 3y ago

I'm not sure: "I drove with the radio blaring so I could not hear the siren" is something to brag about. More sensors == better. And the fact that LiDAR costs have come down two orders of magnitude since Tesla made this decision suggests that some of the economic drivers for such a decision have changed. Part of the reason I've owned FSD since 2017 and yet spent exactly zero minutes in my car in the expressway, in light traffic, in good weather, in the middle of the day reading a book is because of the lack of an entry on the posted timeline saying something like: "Tesla states it will accept liability for any crashes that occur while FSD was engaged." After all these years, it's fair to say: "In the 7 years that Tesla has accepted money from customers for FSD, Tesla still requires human supervision in all cases." So far they have proven they are asmpytotically approaching something that drives badly and is in danger of hitting things.

abenga 3y ago

Why is more sensors less promising?

trabant00 3y ago

My system is even better, it requires no additional hardware whatsoever, nor any single line of code. And it's not only for cars but any vehicle in any environment. It doesn't work for now, but that only means it's in the same state as the Tesla system, meaning very promising.

hnbad 3y ago

This must be a troll comment. Humans don't rely on vision when driving because it's superior, humans rely on vision because that's all we have. Actually that isn't even true as we also use our ears and with assisted driving also the car's various sensors (including LIDAR). The only justification for sticking to vision alone is nerd cred. We don't want human-like AI, we want safe AI. Right now Tesla's AI is making mistakes no human would ever make so it's not even getting that part right. Not using a "crazy number of sensors" is not something you can brag about outside a kind of "sell me this pen" sales pitch.

klabb3 3y ago

I don’t know the specifics and have no dog in the fight, but the argument is flawed here. Human eye sight is in some ways better than small form factor cameras, eg anti-glaring, dynamic range, light sensitivity at night. Also, our vision is quite bad at some stuff, for instance distance estimates have a high error rate. You can do a lot better than that. Tech is not biology, so in order for a computer to not do something stupid, it generally needs different sensors. Some sensors (depth gauge when scuba diving, altimeters for planes, or even gps) are obvious examples of when tech outperforms humans and is safety critical. Just because humans can drive doesn’t mean computers should replicate our “control loop”.

tsimionescu 3y ago
> Also, our vision is quite bad at some stuff, for instance distance estimates have a high error rate. You can do a lot better than that. Our distance estimates are quite good at human scales, especially when compared to vision based approaches. We don't have great precision (we will never say something like "it's 19.7m away") but we very rarely grossly misestimate (think something is 10m away when it's actually 200m away or 10cm away). Of course, compared to a LIDAR or advanced techniques used in physics, we don't hold the tiniest candles (say, if we were to compare to LIGO's distance measurements).