Will Tesla ever do LIDAR?

[heltok]

True, and I think that day is coming sooner than many realize. And don't forget, two camera lenses can do rangefinding also. We have optical rangefinders and they work very well. I have no idea whether Tesla is using the left and right forward facing cameras for stereo imaging but it wouldn't surprise me one bit.

You can do range with one camera and a neural network or one camera comparing to previous frame given vehicle’s velocity from IMU/motion flow. But the error rate will not be centimeter level like the Lidar that measures range natively and there will be very different failure cases.

 

[S4WRXTTCS]

Lidars are awesome and so are cameras. Lidars are active, cameras are passive, lidars measure range, cameras measure light. Both have their usefulness and cost associated with them. For now Tesla has decided that a non-lidar based solution was the best approach. For the future we will see. If they decide to they will likely have done a good calculation of the benefits and costs.

There is a huge benefit for Tesla to keep the same hardware suite over a long time, it makes developing the software so much easier. At some point once vision is ~solved they might decide that, in order to push the systems performance to the next level, adding Lidar might be beneficial, but I don’t think that will be for a few more years at the least.

I guesstimate
1year: <10%
2year: <20%
5year: <50%
10year: >50%

Googles decided to bet heavily into lidar. They could afford to do it as Lidar hardware was not ready and they had to scale slower. But in the long run their solution might be better as they have a longer time to train on the end goal. We will see if early scale beats longer time with more hardware.

I like how you split it up where you looked at it in terms of when they might incorporate lidar. I also liked that you changed it from the name of what they are to what they're actual function is.

Long term (5-10 years) I think it's important to mention that there are Lidar sensors that look just like Camera Sensors. Things like an SPAD array.

So at some point I can easily see an automotive company simply including them as part of the camera assembly house. Like tri-cam plus the SPAD sensor for Lidar.

 

[StealthP3D]

In some ways Camera+Radar+Lidar is a cheat in that it enables higher levels of safety more quickly than trying to do Camera+radar only.

That hasn't been proven yet. It is just a belief you have. Systems that use pre-mapped roads are the quickest way to FSD but Tesla is developing a system than can drive on unmapped roads safely.

We also don't know the level of safety that will be required. I don't believe the 2X better than an average driver will be good enough, and I think 10X is what we should be aiming for. But, I have no idea what the public will find acceptable.

It would be foolish to reject a system that was twice as safe as human drivers because that would be the equivalent of throwing billions of dollars away and killing thousands of people unnecessarily. Safer is safer. Period.

 

[S4WRXTTCS]

It would be foolish to reject a system that was twice as safe as human drivers because that would be the equivalent of throwing billions of dollars away and killing thousands of people unnecessarily. Safer is safer. Period.

Last time I checked humans didn't have this Vulcan way of looking at things.

Instead it's an hysteria level event anytime a machine kills someone.

People are anti-robots.

As someone that loves robots I hate that. But, it is what it is.

Anyways, who do you define as a human driver? Obviously Germany drivers are a lot better than American drivers. So if its 2X as safe as a German driver then I'd be okay with that. If it's 2X a Florida driver then I'm going to have to say no thanks.

 

[S4WRXTTCS]

That hasn't been proven yet. It is just a belief you have.

That's not just my belief.

That the opinion of pretty much all self-driving car experts.

 

[StealthP3D]

That's not just my belief.

That the opinion of pretty much all self-driving car experts.

LOL! Only if you define LIDAR fanboys as "self-driving car experts"! LOL!

 

[electronblue]

@StealthP3D Isn’t it a fact though that nearly all autonomous driving projects are using Lidar at least as a redundancy?

Tesla being a notable exception with maybe a startup or two.

 

[gangzoom]

The 9 eyes don't seem to help it prevent it from crashing into things.

What senses do you use when driving to stop you crashing into things. Also do you think your eyes are better now than compared to when you were 8?

Whats the main difference between a 8 year old and you interms of ability to drive? Is it to do with hardware or software??

The most important part of the human body that is needed to drive is your brain, people drive with one eye, physical disability, but no one can drive if your occipital lobe is damaged.

 

[electronblue]

@gangzoom The argument is, for camera vision to have zero false negatives, it will need to be able to recognize reliably all obstancles and their range from the car. It is not enough to recognize some obstacle types (like cars or people), it will also need to recognize debree and whatnot. (Lacking such detection is one of the biggest obstacles against car responsible driving in the Autopilot today.)

This is where Lidar shines because it has basically zero false negatives and excellent ranging capabilities. It will know something is there, if something is there, and how far away it is. That makes it computationally a great aid for redundancy because it needs so little to do so much. It can even do this in complete darkness.

Now there are other issues with Lidar for sure, like its range being negatively affected by adverse weather, so that is where sensor fusion comes in. As said in all of the recent death-resulting Autopilot scenarios, a Lidar would have reliably seen the obstacle ahead.

The most important part of the human body that is needed to drive is your brain, people drive with one eye, physical disability, but no one can drive if your occipital lobe is damaged.

Indeed. And the issue with cameras is our AI is nowhere near human brain level when it comes to interpreting what it sees. It will get there eventually, but even the arguable leader in camera vision MobilEye is recommending Lidar as redundancy.

 

[StealthP3D]

@StealthP3D Isn’t it a fact though that nearly all autonomous driving projects are using Lidar at least as a redundancy?

I find that kind of argument ridiculous. It's like saying, "Isn't it true that over 98% of all new car sales use an internal combustion engine, at least as a back-up?".

Or during the development of airplanes: "Isn't it true that all flying animals use a flapping motion of their wings to attain flight?". Without flapping wings, the development of the aeroplane is destined to fail.

 

[electronblue]

I find that kind of argument ridiculous. It's like saying, "Isn't it true that over 98% of all new car sales use an internal combustion engine, at least as a back-up?".

I guess fair enough but doesn’t that (at least likely) mean most current self-driving experts indeed are pro Lidar? Maybe the experts are wrong but I don’t think that was the argument...

Or during the development of airplanes: "Isn't it true that all flying animals use a flapping motion of their wings to attain flight?". Without flapping wings, the development of the aeroplane is destined to fail.

Probably a poor analogy given that using cameras is closer to ”flapping wings” (analogous to using eyes like humans or animals) and using Lidar is closer to fixed wings with a non-animal-based engines...

 

[StealthP3D]

I guess fair enough but doesn’t that (at least likely) mean most current self-driving experts indeed are pro Lidar? Maybe the experts are wrong but I don’t think that was the argument...

Maybe most of them are LIDAR fanboys, I don't know because I've never surveyed them. I will note that none of them have succeeded yet. I'm one that believes the self-driving "secret sauce" is in the AI, not the sensors. If the experts at Tesla thought LIDAR would get them to 100% autonomy more quickly, I think they would adopt it in a flash.

I haven't seen them make any moves in that direction (and, frankly, I don't expect to see that either).

 

[fasteddie7]

It must have been kicked around at some point, or it was used to compare and research an optics only approach.

https://www.teslarati.com/tesla-lidar-sensors-spotted-testing-palo-alto/

 

[electronblue]

Maybe most of them are LIDAR fanboys, I don't know because I've never surveyed them

But seriously, though, you can’t really say only Lidar fanboys believe in Lidar. Great many experts in the field seem to believe it is a useful ingredient. Right or wrong, they can’t all be simply fanboys.

 

[StealthP3D]

But seriously, though, you can’t really say only Lidar fanboys believe in Lidar. Great many experts in the field seem to believe it is a useful ingredient. Right or wrong, they can’t all be simply fanboys.

My definition of "LIDAR fanboy" is they believe LIDAR is the best path to autonomous driving. And to be clear, they believe this on faith, not on any solid proof.

 

[electronblue]

My definition of "LIDAR fanboy" is they believe LIDAR is the best path to autonomous driving. And to be clear, they believe this on faith, not on any solid proof.

Fair enough, your opinion is now clear for me I think.

 

[diplomat33]

My definition of "LIDAR fanboy" is they believe LIDAR is the best path to autonomous driving. And to be clear, they believe this on faith, not on any solid proof.

I don't think that is entirely fair. LIDAR does have well known benefits. So it is not really faith based. However, it is also true that since nobody has achieved FSD yet, it is impossible to know for sure what is the "best path" to FSD. Different companies are trying different approaches, trusting in the logic behind their methodology that it is the best.

 

[StealthP3D]

I don't think that is entirely fair. LIDAR does have well known benefits. So it is not really faith based. However, it is also true that since nobody has achieved FSD yet, it is impossible to know for sure what is the "best path" to FSD.

Well, you pretty much confirmed my observation that those who believe LIDAR is the best path to FSD is based on faith by admitting no one has fully solved Full Self Driving. Because there is no solution yet, we don't know who will be the first to solve it. Keep in mind, the successful "solution" will need mass market adoption to be considered solved. A million dollar computer on wheels might be able to self drive but was it really the best path if no one buys it?

 

[Cheburashka]

Well, you pretty much confirmed my observation that those who believe LIDAR is the best path to FSD is based on faith by admitting no one has fully solved Full Self Driving. Because there is no solution yet, we don't know who will be the first to solve it. Keep in mind, the successful "solution" will need mass market adoption to be considered solved. A million dollar computer on wheels might be able to self drive but was it really the best path if no one buys it?

LIDAR has obvious benefits.

Even if you believe that FSD can be solved by vision only, LIDAR is an excellent failsafe.

This is what happens when you rely on vision + radar: Autopilot doesn't detect a truck partially in lane :( : teslamotors

 

[diplomat33]

Well, you pretty much confirmed my observation that those who believe LIDAR is the best path to FSD is based on faith by admitting no one has fully solved Full Self Driving. Because there is no solution yet, we don't know who will be the first to solve it. Keep in mind, the successful "solution" will need mass market adoption to be considered solved. A million dollar computer on wheels might be able to self drive but was it really the best path if no one buys it?

I definitely agree with you on the last part. FSD has to be marketable. My earlier point was just that they do have actual scientific reasons for believing their method is the best. They are not believing in their method based purely on blind faith.