My point stands- drive the car.
You’re 100% right with everything you say. It has to work 100% of the time. Good enough most of the time won’t get to L3-5 self driving.
Camera only is not authorize in most logistic operation in factory, im not sure what changes for a car.
The question is not the camera, it’s what you do with the data that comes off the camera.
The first few versions of camera-based autopilot sucked. They were notably inferior to their radar-based equivalents- that’s because the cameras were using neural network based image recognition on each camera. So it’d take a picture from one camera, say ‘that looks like a car and it looks like it’s about 20’ away’ and repeat this for each frame from each camera. That sorta worked okay most of the time but it got confused a lot. It would also ignore any image it couldn’t classify, which of course was no good because lots of ‘odd’ things can threaten the car. This setup would never get to L3 quality or reliability. It did tons of stupid shit all the time.
What they do now is called occupancy networks. That is, video from ALL cameras is fed into one neural network that understands the geometry of the car and where the cameras are. Using multiple frames of video from multiple cameras at once, it then generates a 3d model of the world around the car and identifies objects in it like what is road and what is curb and sidewalk and other vehicles and pedestrians (and where they are moving and likely to move to), and that data is fed to a planner AI that decides things like where the car should accelerate/brake/turn.
Because the occupancy network is generating a 3d model, you get data that’s equivalent to LiDAR (3d model of space) but with much less cost and complexity. And because you only have one set of sensors, you don’t have to do sensor fusion to resolve discrepancies between different sensors.
I drive a Tesla. And I’m telling you from experience- it DOES work. The latest betas of full self driving software are very very good. On the highway, the computer is a better driver than me in most situations. And on local roads- it navigates them near-perfectly, the only thing it sometimes has trouble with is figuring out when is it’s turn in an intersection (you have to push the gas pedal to force it to go).
I’d say it’s easily at L3+ state for highway driving. Not there yet for local roads. But it gets better with every release.
I know that it can be done :). It’s my direct field of research (localization and mapping of autonomous robots with a focus on building 3D model from camera images e.g NeRF related methods )what i was trying to say is that you cannot have high safety using just cameras. But I think we agree there :)
I’ll be curious to know how they handle environment with a clear lack of depth information (highway roads), how they optimized the processing power (estimating depth is one thing but building a continuous 3D model is different), and the image blur when moving at high speed :). Sensor fusion between visual slam and LiDAR is not complex (since the LiDAR provide what you estimate with your neural occupancy grid anyway, what you get is a more accurate measurement) so on the technological side they don’t really gain much, mainly a gain for the cost.
My guess is that they probably still do a lot of feature detection (lines and stuff) in the background and a lot of what you experience when you drive is improvement in depth estimation and feature detection on rgb images? But maybe not I’ll be really interested to read about it more :). Do you have the research paper that the Tesla algo relies on?
Just to be clear, i have no doubt it works :). I have used similar system for mobile robots and I don’t see why it would not. But I’m also worried they it will lull people in a false sense of safety while the driver should stay alert.
Don’t have the paper, my info comes mainly from various interviews with people involved in the thing. Elon of course, Andrej Karpathy is the other (he was in charge of their AI program for some time).
They apparently used to use feature detection and object recognition in RGB images, then gave up on that (as generating coherent RGB images just adds latency and object recognition was too inflexible) and they’re now just going by raw photon count data from the sensor fed directly into the neural nets that generate the 3d model. Once trained this apparently can do some insane stuff like pull edge data out from below the noise floor.
This may be of interest– This is also from 2 years ago, before Tesla switched to occupancy networks everywhere. I’d say that’s a pretty good equivalent of a LiDAR scan…
I Googled it to see because I thought they maybe were using event cameras then but no, they use 10bit instead of classic 8bit but they are not litterally counting photons (which would not be useful). It’s interesting that it improved the precision and recall of their « object detection model ». Guess the image is of better quality then.
The link from 2 years ago is not particularly impressive: https://arxiv.org/abs/1406.2283 this is an equal valent paper I think from 2014
Not sure the exact details- I heard they were sampling 10 bits per pixel but a bunch of their release notes talked about photon count detection back when they switched to that system.
Given that the HW3 cameras started being used to just generate RGB images, I suspect the current iteration is working by just pulling RAW format frames and interpreting them as a photon count grid, from there detecting edges and geometry with the occupancy network.
I’ve not seen much of anything published by Tesla on the subject. I suspect most of their research they are keeping hush hush to get a leg up on the competition. They share everything regarding EV tech because they want to push the industry in that direction, but I think they see FSD as their secret sauce that they might sell hardware kits but not let others too far under the hood.
Because the occupancy network is generating a 3d model, you get data that’s equivalent to LiDAR (3d model of space) but with much less cost and complexity. And because you only have one set of sensors, you don’t have to do sensor fusion to resolve discrepancies between different sensors.
That’s my problem, it is approximating LIDAR but it isn’t the same. I would say multiple sensor types is necessary for exactly the reason you suggested it isn’t - to get multiple forms of input and get consensus, or failing consensus fail-safe.
I don’t doubt Tesla autopilot works well and it certainly seems to be an impressive feat of engineering, but can it be better?
In our town we had a Tesla shoot through red traffic lights near our local school barely missing a child crossing the road. The driver was looking at their lap (presumably their phone). I looked online and apparently autopilot doesn’t work with traffic lights, but FSD does?
It’s not specific to Tesla but people unaware of the limitations level 2, particularly when brands like Tesla give people the impression the car “drives itself” is unethical.
My opinion is if that Tesla had extra sensors, even if the car is only in level 2 mode, it should be able to pick up that something is there and slow/stop. I want the extra sensors to cover the edge cases and give more confidence in the system.
Would you still feel the same about Tesla if your car injured/killed someone or if someone you care about was injured/killed by a Tesla?
IMHO these are not systems that we should be compromising to cut costs or because the CEO is too stubborn. If we can put extra sensors in and it objectively makes it safer why don’t we? Self driving cars are a luxury.
Crazy hypothetical: I wonder how Tesla would cope with someone/something covered in Vantablack?
In our town we had a Tesla shoot through red traffic lights near our local school barely missing a child crossing the road. The driver was looking at their lap (presumably their phone). I looked online and apparently autopilot doesn’t work with traffic lights, but FSD does?
There’s a few versions of this and several generations with different capability. The early Tesla Autopilot had no recognition of stop signs, it was literally just ‘cruise control that keeps you in your lane’. FSD for sure does recognize stop signs, traffic lights, etc and reacts correctly to them. I BELIEVE that the current iteration of Traffic Aware Cruise Control (what you get if you don’t pay extra for FSD or Enhanced Autopilot) will stop for traffic lights but I could be wrong on that. I know it detects pedestrians but its detection isn’t nearly as advanced as FSD.
I will give you that in theory, the time-of-flight data from a LiDAR pulse will give you a more reliable point cloud than anything you’d get from cameras. But I also know Tesla is doing things with cameras that border on black magic. They gave up on getting images out of the cameras and are now just using the raw photon count data from the sensor, and with the AI trained it can apparently detect edges with only a few photons of difference between pixels (below the noise floor). And I can say from experience that a few times I’ve been in blackout rainstorms where even with full wipers I can barely see anything, and the FSD visualization doesn’t skip a beat and it sees other cars before I do.
Would you still feel the same about Tesla if your car injured/killed someone or if someone you care about was injured/killed by a Tesla?
As a Level 2 system, the Tesla is not capable of injuring or killing someone. The driver is responsible for that.
But I’d ask- if a Tesla saw YOUR loved one in the road, and it would have reacted but it wasn’t in FSD mode and the human driver reacted too slowly, how would you feel about that? I say this not to be contrarian, but because we really are approaching the point where the car has better situational awareness than the human.
If we can put extra sensors in and it objectively makes it safer why don’t we? Self driving cars are a luxury.
For the reason above with the loved one. If you can use cameras and make a system that costs the manufacturer $3000/car, and it’s 50 times safer than a human, or use LiDAR and cost the manufacturer $10,000/car, and it’s 100 times safer than a human, which is safer?
The answer is the cameras, because it will be on more cars, thus deliver more overall safety.
I understand the thinking that ‘Elon cheaped out, Tesla FSD is a hack system on shitty hardware that uses clever programming to work around a cut-rate sensor suite’. But I’d also argue- if they can get similar performance out of a camera, and put it on more cars, doesn’t that do more to overall improve safety?
In the example above, if the car didn’t have the self driving package because the guy couldn’t afford it, wouldn’t you prefer that a decent but better than human self driving system was on the car?
There’s a few versions of this and several generations with different capability. […]
This raises its own issues but is the nature of the “move fast and break things” ethos of tech today. While it has its benefits; is it suitable for vehicles, particularly their safety systems? It isn’t clear to me, as it is a double-edged sword.
As a Level 2 system, the Tesla is not capable of injuring or killing someone. The driver is responsible for that.
But I’d ask- if a Tesla saw YOUR loved one in the road, and it would have reacted but it wasn’t in FSD mode and the human driver reacted too slowly, how would you feel about that? I say this not to be contrarian, but because we really are approaching the point where the car has better situational awareness than the human.
I would be angry that such a modern car with any form of self driving doesn’t have emergency braking. Though, that would require additional sensors…
I’d also be angry that L2 systems were allowed in that environment in the first place, but as you say it is ultimately the drivers fault.
Like cruise control having minimum speeds that generally prevent it being used in town; I would hope that the manufacturer would make it difficult to use L2 outside of motorway driving. This doesn’t prevent people bypassing it but means someone doing so was trying to do something they shouldn’t.
With a connected vehicle, being able to limit L2 use outside of motorway should be straightforward.
Then it becomes akin to disabling traction control or adaptive cruise control and having an accident that could be prevented. The tools are there, the default is on, a driver deliberately disabled it. The manufacturer did as much as they reasonably could.
In the example above, if the car didn’t have the self driving package because the guy couldn’t afford it, wouldn’t you prefer that a decent but better than human self driving system was on the car?
I would prefer they had no self driving rather than be under the mistaken impression the car could drive for them in the current configuration. The limitations of self driving (in any car) are often not clear to a lot of people and can vary greatly. I feel this is where accidents are most likely - in the stage between fully manual and fully autonomous.
If Tesla offer a half-way for less money would you not expect the consumer to take the cheapest option? If they have an accident it is more likely someone else is injured, so why pay more to improve the self driving when it doesn’t affect them?
If you can use cameras and make a system that costs the manufacturer $3000/car, and it’s 50 times safer than a human, or use LiDAR and cost the manufacturer $10,000/car, and it’s 100 times safer than a human, which is safer?
The answer is the cameras, because it will be on more cars, thus deliver more overall safety.
I agree an improvement is better than none, but I’m not sure your conclusion can be made so easily? Tesla is the only company I know steadfastly refusing to use any other sensor types and the only reason I see is price.
Thinking about it, drum brakes are cheaper than disc brakes… (said with tongue-firmly-in-cheek)
Another concern is that any Tesla incidents, however rare, could do huge damage to people’s perception of self driving. People mightn’t know there is a difference between Tesla and other manufacturer’s autonomous driving ability.
For many people Tesla is self-driving cars, if a Tesla has an accident in L2 even though this is the driver’s fault the headlines will be “Tesla autopilot hits school child” not “Driver inappropriately uses limited motorway assistance mode of car in small town hitting school child”
What about the impact on the industry? If Tesla is much cheaper than LIDAR-equipped vehicles will this kill a better/safer product a-la betamax?
IMHO safety shouldn’t take a lower priority to price/CEO demands. Consumers often don’t know and frankly shouldn’t need to know the details of these systems.
Do you pick your airline based on the plane they fly and it’s safety record or the price of the ticket, being confident all aviation is held to rigorous safety standards?
As has been seen recently with a certain submarine, safety measures should not be taken lightly.
My point stands- drive the car.
You’re 100% right with everything you say. It has to work 100% of the time. Good enough most of the time won’t get to L3-5 self driving.
The question is not the camera, it’s what you do with the data that comes off the camera.
The first few versions of camera-based autopilot sucked. They were notably inferior to their radar-based equivalents- that’s because the cameras were using neural network based image recognition on each camera. So it’d take a picture from one camera, say ‘that looks like a car and it looks like it’s about 20’ away’ and repeat this for each frame from each camera. That sorta worked okay most of the time but it got confused a lot. It would also ignore any image it couldn’t classify, which of course was no good because lots of ‘odd’ things can threaten the car. This setup would never get to L3 quality or reliability. It did tons of stupid shit all the time.
What they do now is called occupancy networks. That is, video from ALL cameras is fed into one neural network that understands the geometry of the car and where the cameras are. Using multiple frames of video from multiple cameras at once, it then generates a 3d model of the world around the car and identifies objects in it like what is road and what is curb and sidewalk and other vehicles and pedestrians (and where they are moving and likely to move to), and that data is fed to a planner AI that decides things like where the car should accelerate/brake/turn.
Because the occupancy network is generating a 3d model, you get data that’s equivalent to LiDAR (3d model of space) but with much less cost and complexity. And because you only have one set of sensors, you don’t have to do sensor fusion to resolve discrepancies between different sensors.
I drive a Tesla. And I’m telling you from experience- it DOES work. The latest betas of full self driving software are very very good. On the highway, the computer is a better driver than me in most situations. And on local roads- it navigates them near-perfectly, the only thing it sometimes has trouble with is figuring out when is it’s turn in an intersection (you have to push the gas pedal to force it to go).
I’d say it’s easily at L3+ state for highway driving. Not there yet for local roads. But it gets better with every release.
It’s an interesting discussion thanks!
I know that it can be done :). It’s my direct field of research (localization and mapping of autonomous robots with a focus on building 3D model from camera images e.g NeRF related methods )what i was trying to say is that you cannot have high safety using just cameras. But I think we agree there :)
I’ll be curious to know how they handle environment with a clear lack of depth information (highway roads), how they optimized the processing power (estimating depth is one thing but building a continuous 3D model is different), and the image blur when moving at high speed :). Sensor fusion between visual slam and LiDAR is not complex (since the LiDAR provide what you estimate with your neural occupancy grid anyway, what you get is a more accurate measurement) so on the technological side they don’t really gain much, mainly a gain for the cost.
My guess is that they probably still do a lot of feature detection (lines and stuff) in the background and a lot of what you experience when you drive is improvement in depth estimation and feature detection on rgb images? But maybe not I’ll be really interested to read about it more :). Do you have the research paper that the Tesla algo relies on?
Just to be clear, i have no doubt it works :). I have used similar system for mobile robots and I don’t see why it would not. But I’m also worried they it will lull people in a false sense of safety while the driver should stay alert.
Don’t have the paper, my info comes mainly from various interviews with people involved in the thing. Elon of course, Andrej Karpathy is the other (he was in charge of their AI program for some time).
They apparently used to use feature detection and object recognition in RGB images, then gave up on that (as generating coherent RGB images just adds latency and object recognition was too inflexible) and they’re now just going by raw photon count data from the sensor fed directly into the neural nets that generate the 3d model. Once trained this apparently can do some insane stuff like pull edge data out from below the noise floor.
This may be of interest– This is also from 2 years ago, before Tesla switched to occupancy networks everywhere. I’d say that’s a pretty good equivalent of a LiDAR scan…
I Googled it to see because I thought they maybe were using event cameras then but no, they use 10bit instead of classic 8bit but they are not litterally counting photons (which would not be useful). It’s interesting that it improved the precision and recall of their « object detection model ». Guess the image is of better quality then.
The link from 2 years ago is not particularly impressive: https://arxiv.org/abs/1406.2283 this is an equal valent paper I think from 2014
Not sure the exact details- I heard they were sampling 10 bits per pixel but a bunch of their release notes talked about photon count detection back when they switched to that system.
Given that the HW3 cameras started being used to just generate RGB images, I suspect the current iteration is working by just pulling RAW format frames and interpreting them as a photon count grid, from there detecting edges and geometry with the occupancy network.
I’ve not seen much of anything published by Tesla on the subject. I suspect most of their research they are keeping hush hush to get a leg up on the competition. They share everything regarding EV tech because they want to push the industry in that direction, but I think they see FSD as their secret sauce that they might sell hardware kits but not let others too far under the hood.
I think you are absolutely correct for the interpretation of the photon count :)
That’s my problem, it is approximating LIDAR but it isn’t the same. I would say multiple sensor types is necessary for exactly the reason you suggested it isn’t - to get multiple forms of input and get consensus, or failing consensus fail-safe.
I don’t doubt Tesla autopilot works well and it certainly seems to be an impressive feat of engineering, but can it be better?
In our town we had a Tesla shoot through red traffic lights near our local school barely missing a child crossing the road. The driver was looking at their lap (presumably their phone). I looked online and apparently autopilot doesn’t work with traffic lights, but FSD does?
It’s not specific to Tesla but people unaware of the limitations level 2, particularly when brands like Tesla give people the impression the car “drives itself” is unethical.
My opinion is if that Tesla had extra sensors, even if the car is only in level 2 mode, it should be able to pick up that something is there and slow/stop. I want the extra sensors to cover the edge cases and give more confidence in the system.
Would you still feel the same about Tesla if your car injured/killed someone or if someone you care about was injured/killed by a Tesla?
IMHO these are not systems that we should be compromising to cut costs or because the CEO is too stubborn. If we can put extra sensors in and it objectively makes it safer why don’t we? Self driving cars are a luxury.
Crazy hypothetical: I wonder how Tesla would cope with someone/something covered in Vantablack?
There’s a few versions of this and several generations with different capability. The early Tesla Autopilot had no recognition of stop signs, it was literally just ‘cruise control that keeps you in your lane’. FSD for sure does recognize stop signs, traffic lights, etc and reacts correctly to them. I BELIEVE that the current iteration of Traffic Aware Cruise Control (what you get if you don’t pay extra for FSD or Enhanced Autopilot) will stop for traffic lights but I could be wrong on that. I know it detects pedestrians but its detection isn’t nearly as advanced as FSD.
I will give you that in theory, the time-of-flight data from a LiDAR pulse will give you a more reliable point cloud than anything you’d get from cameras. But I also know Tesla is doing things with cameras that border on black magic. They gave up on getting images out of the cameras and are now just using the raw photon count data from the sensor, and with the AI trained it can apparently detect edges with only a few photons of difference between pixels (below the noise floor). And I can say from experience that a few times I’ve been in blackout rainstorms where even with full wipers I can barely see anything, and the FSD visualization doesn’t skip a beat and it sees other cars before I do.
As a Level 2 system, the Tesla is not capable of injuring or killing someone. The driver is responsible for that.
But I’d ask- if a Tesla saw YOUR loved one in the road, and it would have reacted but it wasn’t in FSD mode and the human driver reacted too slowly, how would you feel about that? I say this not to be contrarian, but because we really are approaching the point where the car has better situational awareness than the human.
For the reason above with the loved one. If you can use cameras and make a system that costs the manufacturer $3000/car, and it’s 50 times safer than a human, or use LiDAR and cost the manufacturer $10,000/car, and it’s 100 times safer than a human, which is safer?
The answer is the cameras, because it will be on more cars, thus deliver more overall safety.
I understand the thinking that ‘Elon cheaped out, Tesla FSD is a hack system on shitty hardware that uses clever programming to work around a cut-rate sensor suite’. But I’d also argue- if they can get similar performance out of a camera, and put it on more cars, doesn’t that do more to overall improve safety?
In the example above, if the car didn’t have the self driving package because the guy couldn’t afford it, wouldn’t you prefer that a decent but better than human self driving system was on the car?
This raises its own issues but is the nature of the “move fast and break things” ethos of tech today. While it has its benefits; is it suitable for vehicles, particularly their safety systems? It isn’t clear to me, as it is a double-edged sword.
I would be angry that such a modern car with any form of self driving doesn’t have emergency braking. Though, that would require additional sensors…
I’d also be angry that L2 systems were allowed in that environment in the first place, but as you say it is ultimately the drivers fault.
Like cruise control having minimum speeds that generally prevent it being used in town; I would hope that the manufacturer would make it difficult to use L2 outside of motorway driving. This doesn’t prevent people bypassing it but means someone doing so was trying to do something they shouldn’t.
With a connected vehicle, being able to limit L2 use outside of motorway should be straightforward.
Then it becomes akin to disabling traction control or adaptive cruise control and having an accident that could be prevented. The tools are there, the default is on, a driver deliberately disabled it. The manufacturer did as much as they reasonably could.
I would prefer they had no self driving rather than be under the mistaken impression the car could drive for them in the current configuration. The limitations of self driving (in any car) are often not clear to a lot of people and can vary greatly. I feel this is where accidents are most likely - in the stage between fully manual and fully autonomous.
If Tesla offer a half-way for less money would you not expect the consumer to take the cheapest option? If they have an accident it is more likely someone else is injured, so why pay more to improve the self driving when it doesn’t affect them?
I agree an improvement is better than none, but I’m not sure your conclusion can be made so easily? Tesla is the only company I know steadfastly refusing to use any other sensor types and the only reason I see is price.
Thinking about it, drum brakes are cheaper than disc brakes… (said with tongue-firmly-in-cheek)
Another concern is that any Tesla incidents, however rare, could do huge damage to people’s perception of self driving. People mightn’t know there is a difference between Tesla and other manufacturer’s autonomous driving ability.
For many people Tesla is self-driving cars, if a Tesla has an accident in L2 even though this is the driver’s fault the headlines will be “Tesla autopilot hits school child” not “Driver inappropriately uses limited motorway assistance mode of car in small town hitting school child”
What about the impact on the industry? If Tesla is much cheaper than LIDAR-equipped vehicles will this kill a better/safer product a-la betamax?
IMHO safety shouldn’t take a lower priority to price/CEO demands. Consumers often don’t know and frankly shouldn’t need to know the details of these systems.
Do you pick your airline based on the plane they fly and it’s safety record or the price of the ticket, being confident all aviation is held to rigorous safety standards?
As has been seen recently with a certain submarine, safety measures should not be taken lightly.