Waymo brings in $2.25 billion from outside investors, Alphabet

[diplomat33]

The only reason I even post here, is because of you "total BS" about need for lidar.

It's not BS. I've already explained why lidar is needed.

But now, you are writing software, to do vision specific to Lidar, so instead of focusing your energy on the root problem of vision, you are fragmenting your team into solving the Lidar vision problems and camera vision.

You are taking the Tesla view that camera vision alone is all you need so why bother with lidar vision too. But you need lidar vision so it makes sense to work on lidar vision and camera vision.

 

[diplomat33]

You end up literally writing vision for lidar...

Yes, you need to write vision for lidar. You need vision for both lidar and camera. Camera vision alone is not enough.

 

[diplomat33]

@mspisars Basically, this entire debate revolves around the question of whether camera vision alone is enough. If camera vision alone is good enough then no, you don't need lidar vision (if you don't care about sensor redundancy). That's the Tesla approach: cameras give you all the info you need for driving so you just need camera vision. But if camera vision is not good enough, then you need lidar vision.

 

[VT_EE]

LIDAR will work in light rain and light snow. Cameras will fail in heavy rain and snow too. Just think about your vision. How well do you see in heavy rain, dense fog or snow. Answer: not well.

Cameras don't work well at night whereas lidar works great at night. Radar is excellent in rain and snow.

That is why you need cameras, radar and lidar together. When you combine all three sensors together, each sensor will compensate when another sensor can't see well, and you will get a good 3D map of the world in all weather conditions

Actually the cameras do an excellent job during heavy rain. The system detects road lines better than I do. Given the problems with laser scattering on rain droplets and snow, I am highly skeptical LIDAR will function well in any sort of moderately bad weather, which means the system is back to using radar and cameras.

 

[VT_EE]

@mspisars Basically, this entire debate revolves around the question of whether camera vision alone is enough. If camera vision alone is good enough then no, you don't need lidar vision (if you don't care about sensor redundancy). That's the Tesla approach: cameras give you all the info you need for driving so you just need camera vision. But if camera vision is not good enough, then you need lidar vision.

If camera vision is not enough, then none of us should be driving.

 

[mikes_fsd]

if you don't care about sensor redundancy

Tesla provides sensor redundancy, by using multiple overlapping cameras. For some reason you want redundancy to mean to use a completely different sensor suite.

 

[diplomat33]

If camera vision is not enough, then none of us should be driving.

Human vision+brain is good enough for driving. We have a complex NN in our brain that understands the world perfectly. We don't have computer vision that is that good yet. That is what Tesla is working on.

 

[diplomat33]

Tesla provides sensor redundancy, by using multiple overlapping cameras. For some reason you want redundancy to mean to use a completely different sensor suite.

Yes, true redundancy means different sensors. Overlapping cameras is good but only provides redundancy where there is overlap.

 

[mikes_fsd]

Yes, true redundancy means different sensors. Overlapping cameras is good but only provides redundancy where there is overlap.

I would not like to see the monstrosity of a FSD car that you would create

Forward facing you have three cameras that overlap 100% plus two b pillar cameras that give 50% each overlap to the front view

 

[diplomat33]

My final word on the subject:

Lidar allows you to get to FSD faster and more reliably. Lidar will tell you where everything is. You still need computer vision but it does not need to be perfect to start doing FSD. You can solve perception faster with lidar and then move on to writing the driving policy sooner.

Obviously, Tesla feels that since you need computer vision regardless, you might as well, just work on perfect computer vision and then you won't need lidar. Fine. But you need perfect computer vision for Tesla's method to work. If your computer vision makes a mistake, you're screwed. Perfect computer vision is going to be a lot harder than using lidar. So it will take a lot longer to get to FSD.

 

[diplomat33]

I would not like to see the monstrosity of a FSD car that you would create.

You mean like this?

Forward facing you have three cameras that overlap 100% plus two b pillar cameras that give 50% each overlap to the front view

That's overlap in front. What about in the rear or sides?

 

[wuxyjj]

Thanks for all the information. This is the future and I am very interested to learn. I am a newbie on this but the common sense tell me one needs two critical components to make the AD happen: Sensor system to get information (hardware) and then the AI (software) to tell car how to drive based on the information. For me, the software will be the key. As for the sensor system, more accurate information it can gather, the better. So I do not understand why Tesla insists on cameras only, unless it adds on too much cost otherwise. The question is, why Waymo can do L4 now but Tesla cannot? Is it because the hardware or the software? If it is just software, then Tesla should be able to catch up with all the data gathered from all those AP drivings. If Waymo can do it, why cannot Tesla?

 

[Daniel in SD]

Tesla's computer vision could not recognize this as an obstacle last year. Computer vision is much harder than you think.

 

[Daniel in SD]

Thanks for all the information. This is the future and I am very interested to learn. I am a newbie on this but the common sense tell me one needs two critical components to make the AD happen: Sensor system to get information (hardware) and then the AI (software) to tell car how to drive based on the information. For me, the software will be the key. As for the sensor system, more accurate information it can gather, the better. So I do not understand why Tesla insists on cameras only, unless it adds on too much cost otherwise. The question is, why Waymo can do L4 now but Tesla cannot? Is it because the hardware or the software? If it is just software, then Tesla should be able to catch up with all the data gathered from all those AP drivings. If Waymo can do it, why cannot Tesla?

The software is by far the most difficult part of the problem. What would be the point of shipping expensive hardware when you're not even sure if it will be good enough until you write the software? Waymo is making prototype vehicles while Tesla is making production vehicles. Tesla's approach is extremely unlikely to work but it's not really costing them any extra since you need all the cameras and compute hardware to do driver assist features anyway.
While Waymo is operating a very limited number of L4 vehicles without a safety driver it's certainly not clear yet that they're as good as a human driver and ready to released in more markets.

 

[MP3Mike]

I found out about it elsewhere and found someone had already posted it in the very long TSLA investor thread which I don't read anyway, I figure it make sense in this forum too.
Waymo brings in $2.25 billion from outside investors, Alphabet – TechCrunch
Waypoint - The official Waymo blog: Waymo raises first external investment round

I wonder if each of the players (e.g. Waymo, Cruise, Baidu, Apple, Plus AI, Nuro, etc.) in the autonomous space should each get their own thread w/a new one created when we see news from them?

Is this WaymoGoogle's way of saying we are tired of usingwasting our own money, can you please give us some of yours to burn?

 

[Tam]

...If Waymo can do it, why cannot Tesla?

I think there's a confusion between Waymo's and Tesla's methods.

Based on a method, it can do something well and it can not do something either.

That means, if you want Waymo, you need to sell your house and move to Chandler, AZ and enroll in its rider program.

That means I cannot own and I cannot use Waymo in my own town.

For practical purposes, Waymo can't do what Tesla can in my own town.

So it is misleading to say that "Waymo can do it". No, it can't. Maybe someday it will, but currently, it can't!

It's just like Waymo's elevator works very well in a building with 9 floors, but it hasn't programmed for a building with 10 floors so you have to wait for it to come to your town.

In the meantime, Tesla's elevator can work with a building with any number of floors. You can install it in your own town right now without waiting!

 

[Daniel in SD]

I think there's a confusion between Waymo's and Tesla's methods.

Based on a method, it can do something well and it can not do something either.

That means, if you want Waymo, you need to sell your house and move to Chandler, AZ and enroll in its rider program.

That means I cannot own and I cannot use Waymo in my own town.

For practical purposes, Waymo can't do what Tesla can in my own town.

So it is misleading to say that "Waymo can do it". No, it can't. Maybe someday it will, but currently, it can't!

It's just like Waymo's elevator works very well in a building with 9 floors, but it hasn't programmed for a building with 10 floors so you have to wait for it to come to your town.

In the meantime, Tesla's elevator can work with a building with any number of floors. You can install it in your own town right now without waiting!

Tesla doesn't have autonomous driving anywhere and there isn't much evidence that they're close. They started development in 2016 and they're doing a major code rewrite now. I'm beginning to think L4 capability by the end of this year is not going to happen. When do you think they'll have an autonomous vehicle?

 

[diplomat33]

For practical purposes, Waymo can't do what Tesla can in my own town.

But the reverse is also true. If I go to Chandler, AZ, I can ride a Waymo robotaxi, but I can't ride in a Tesla robotaxi. So Tesla can't do what Waymo can do.

 

[Tam]

Tesla doesn't have autonomous driving anywhere and there isn't much evidence that they're close. They started development in 2016 and they're doing a major code rewrite now. I'm beginning to think L4 capability by the end of this year is not going to happen. When do you think they'll have an autonomous vehicle?

True. There's no question about the above facts at all.

Tesla's system still needs human driver to assist it while Waymo can now have no human driver at the wheel at all in Chandler, AZ.

It's a trade-off between Tesla's L2 generalized solution versus Waymo's geofenced L4 system.

 

[Tam]

But the reverse is also true. If I go to Chandler, AZ, I can ride a Waymo robotaxi, but I can't ride in a Tesla robotaxi. So Tesla can't do what Waymo can do.

Very true indeed!

It's just like a dog can find its way back very well in its own training ground with all familiar environment but the owner refuses to send the dog to a campground because it's not familiar.