Waymo will beat Tesla FSD because of LIDAR

[Crissa]

Yea it's more difficult, but it's more accurate.

That depends.

Simple radar sees pulsed returning from surfaces - but surfaces can be anything from air or objects; and different objects will return a different density of return. So an overhead sign and a bridge look exactly the same to radar. And a flock of insects or birds looks like a rainshower. If you've used a fishfinder with raw output, this is what that looks like. And sometimes you're looking at rocks deep in the mud instead of fish.

The next kind of radar only sees the difference in speed between two objects. It does it because waves reflecting off an object will have their signals slightly changed in the reflection. But the problem here is that things which are going at the same speed as the background look the same as not being there at all. So a stopped car looks the same as the road.

The most recent 'high definition' radar can actually see more than distance and speed. It gives you shapes. That's useful! But it's also expensive. And... Is it really giving more and new data than the many cameras you already need, being used for parallax?

-Crissa

Sponsored

 

[scottf200]

However, multiple cameras will provide depth, similar to how your eyes work by providing triangulated vision. Notice if you close one, eye, you lose distance measuring to a strong degree. But when you use two eyes, you can get back the sense of depth.

Humans typically use binocular vision (two eyes) to perceive depth up to about 15 fee. Beyond this distance, the brain starts relying more on monocular cues (like size, texture, and perspective) to understand depth. I know one-eyed folks well and they understand depth at various distances. Their brains uses many techniques. Tesla's cameras are even closer together than the human eyes

 

[jookyone]

I wrote that intelligence doesn't have to be at the level of a human being. Oxford is defining a field of study, a discipline. Intelligence is intelligence no matter how it's created. It's the ability to learn, understand, and think logically about things, according to Oxford. Are squirrels intelligent? Watch this and tell me.

You are off on a philosophical tangent. I said "there is no AI. fight me" and you are sending me squirrel videos.

You lost. There is no AI.

 

[Cybergirl]

You are off on a philosophical tangent. I said "there is no AI. fight me" and you are sending me squirrel videos.

You lost. There is no AI.

Uncle!

 

[jookyone]

Uncle!

Not to say that objectively, ending a debate with a Mark Rober squirrel video was the correct call and I should have stopped. But I stand by my statements.

 

[REM]

The MIT course came to a stunning conclusion about AI, that machines + humans collaborating in a "supermind" outperform both AI or humans separately. A supermind is actually classified as artificial intelligence, even though it is a human-machine collective

Most people in the know call this "the singularity".

 

[Deleted member 17810]

That depends.

Simple radar sees pulsed returning from surfaces - but surfaces can be anything from air or objects; and different objects will return a different density of return. So an overhead sign and a bridge look exactly the same to radar. And a flock of insects or birds looks like a rainshower. If you've used a fishfinder with raw output, this is what that looks like. And sometimes you're looking at rocks deep in the mud instead of fish.

The next kind of radar only sees the difference in speed between two objects. It does it because waves reflecting off an object will have their signals slightly changed in the reflection. But the problem here is that things which are going at the same speed as the background look the same as not being there at all. So a stopped car looks the same as the road.

The most recent 'high definition' radar can actually see more than distance and speed. It gives you shapes. That's useful! But it's also expensive. And... Is it really giving more and new data than the many cameras you already need, being used for parallax?

-Crissa

Yes, lidar gives accurate, mathematical, distance data, that fsd optical system has proven to not been able to deduct.

 

[Cybergirl]

Not to say that objectively, ending a debate with a Mark Rober squirrel video was the correct call and I should have stopped. But I stand by my statements.

That's fine. I'll stand by mine.

 

[REM]

A few people touched on this, but the absolute fundamental limit of any Ai model is that it HAS to be in human readable format. Real AGI will hit an absolute brick wall when it breaks human observation limits because we simply cannot quantify the data and make it useful.

People don't have lidar; we have eyes (cameras). No amount of sophisticated programming will allow us to process that data to the fullest extent and in a way that makes sense to us. This is the reason every competent researcher tries to explain that lidar introduces too much noise to a system. This is also why Radar is helpful in certain applications, but it mostly becomes too much noise in driver based system. Humans don't have radar; ergo it's mostly noise to us.

 

[mark555055c]

Lots of data and interesting info, but, to put it bluntly, wake me up when there is an attractive LIDAR autonomous vehicle that works anywhere, and that saves me time, money, or stress.

At this moment, Tesla is the only company delivering autonomous vehicle tech that makes commuting easier and more enjoyable.

 

[Deleted member 17810]

A few people touched on this, but the absolute fundamental limit of any Ai model is that it HAS to be in human readable format. Real AGI will hit an absolute brick wall when it breaks human observation limits because we simply cannot quantify the data and make it useful.

People don't have lidar; we have eyes (cameras). No amount of sophisticated programming will allow us to process that data to the fullest extent and in a way that makes sense to us. This is the reason every competent researcher tried to explain that lidar introduces too much noise to a system. This is also why Radar is helpful in certain applications, but it mostly becomes too much noise in driver based system. Humans don't have radar; ergo it's mostly noise to us.

It's noise to "us" but not to the parking system that's about to hit a pillar.

The ENTIRE point of a FSD system is to be better than what "humans" can do.

What the car does at 80 mphs does NOT need to be in human readible format because it's called traction control.

FSD has to be past what people can do.

musk is using the noise as an excuse for lowering costs, not making a better system .

 

[REM]

It's noise to "us" but not to the parking system that's about to hit a pillar.

The ENTIRE point of a FSD system is to be better than what "humans" can do.

What the car does at 80 mphs does NOT need to be in human readible format because it's called traction control.

FSD has to be past what people can do.

musk is using the noise as an excuse for lowering costs, not making a better system .

To further illustrate, what does the programmer do in order to make the car "understand" that it is about to hit a pole?

The programmer must modulate the data pulled from the lidar and transform it into something visual so he can interpret it himself, THEN he has to relay that back to the machine in a format it understands. It's a futile loop with far too many holes.

Every single derivative must be human retable. Otherwise how do we even understand the data?

I think maybe you are confusing the ability to parse the data vs the fact that machines can do certain tasks much faster than we can.

 

[Outdoors]

But, with radar, it could be better in some situations, fog, rain, dark.
I hate that Tesla removed the radar, if you’re going to have a robot do human tasks, why limit it to human capabilities

The radar returns are horrible. Sure improve the radar. Then one is presented with two data sets that fight with each other.

Some may dislike vision. Yet for me it sees the things I can't. Many a time I have had my car do what I thought was phantom braking. Only to see a deer on the other side of the road. A deer I never saw. In the dark I might add.

Can get lost in the weeds on this stuff. I like real world examples from people that drive 50k+ a year on AP/FSD.

 

[Deleted member 17810]

To further illustrate, what does the programmer do in order to make the car "understand" that it is about to hit a pole?

The programmer must modulate the data pulled from the lidar and transform it into something visual so he can interpret it himself, THEN he has to relay that back to the machine in a format it understands. It's a futile loop with far too many holes.

Every single derivative must be human retable. Otherwise how do we even understand the data?

I think maybe you are confusing the ability to parse the data vs the fact that machines can do certain tasks much faster than we can.

I'm not confusing it. I'm saying just that. The system can poll interpret and action data faster than a human can.

The "programmer" is AI that's supposed to be faster and better at manipulating data.

Every single derivative doesn't need to be human interpretable. It needs to be the opposite.

I can't tell where studs are, but a stud finder can tell me it within cms.

 

[dalton108]

Can get lost in the weeds on this stuff. I like real world examples from people that drive 50k+ a year on AP/FSD.

Agree and I’ll tell you who’s been consistently untrustworthy/unreliable on analyzing these issues: YouTubers, automotive reporters and folks who work for/on/with alternate technologies!

Talk to me AFTER you’ve used the technology for longer than just a sufficient amount of time to confirm your biases. ?

Sponsored