I can't see this without immediately thinking of Factorio

[Martoon]

[LordFedora]

I like the reverse of that gif better

"ok, i'm gunna put this block right here, DON'T MESS IT UP THIS TIME"

"lawl, u dn't tell me wut 2 do"

">.>"

[sillyfly]

What do you think that thing at the rightmost end does?

Neat stacks go in, randomized pieces come out, then the two robot-thingies stack them up nicely again

[Martoon]

I like the reverse of that gif better

"ok, i'm gunna put this block right here, DON'T MESS IT UP THIS TIME"

"lawl, u dn't tell me wut 2 do"

">.>"

Ha! It's like L00-C and ETH-L in the chocolate factory.

[cube]

Does it remind anyone else of sushi rolls? I WANT A SUSHI ROBOT!

[DerivePi]

My first thought, "there has got to be a better way."
Second thought, "how would you program the right robot and what kind of optimization is it performing?"
- Looks like it identifies a clear zone with one element (the next one on the belt) already in its correct position
- Then it moves the other 3 based on shortest distance

If we can do this, we should be able to have robots do dishes by now.

[Cilya]

If we can do this, we should be able to have robots do dishes by now.

Genericity is hard to achieve. You would be able to build such a robot if all your plates were identical. Any difference could prevent the robot from 1/ grabbing the plate firmly 2/ clean it appropriately 3/ put it in the right place. We are still doing progress in this matter.

[Rage]

If we can do this, we should be able to have robots do dishes by now.

Genericity is hard to achieve. You would be able to build such a robot if all your plates were identical. Any difference could prevent the robot from 1/ grabbing the plate firmly 2/ clean it appropriately 3/ put it in the right place. We are still doing progress in this matter.

You would need something that could quickly and accurately scan the dishes it's grabbing for not only shape and size but also material. It would then need to run an algorithm that takes into account the size, shape, and material of the dish, the size and shape of the grabber, and the way it actually applies gripping force to determine how and where to grab. Ideally it would determine two points to grab that are fairly close the the center of mass of the object so as to be able to clean where it grabbed the first time and also help prevent any leverage issues. From there you already have the data necessary to effectively clean the dish and you just have to make sure you put it on a drying rack, or cabinet/drawer gently enough. To be honest I believe we already have the technology and software to do it, someone just has to put it together and debug it.

The main for it to be economically viable is that it needs to be better than a human and at lower cost, it doesn't need to be perfect, just better. The hardest part is gonna be the grabbing, after that everything is relatively easy, and something like this would really just be a robot that grabs the dishes, puts the excess food in the trash, puts the dishes in a dish washer, and then once they're clean, puts them away.

[immibis]

If we can do this, we should be able to have robots do dishes by now.

We can. These robots are called dishwashers.

[Cilya]

To be honest I believe we already have the technology and software to do it

I agree, and I believe we have it from a long time ago. The reason why it may not have been done is, like you said, for economic reason. And what is changing currently, is the need of more powerful robots like the ones in this video. I believe the robots in this video are a demonstration of nowdays improvement to automate the last repetitive tasks we didn't automated already. This was probably not economically interesting ten years ago, but is now : if we already automated everything which is simple to automate, the only way to further improve productivity is to complexity these robots by combining all we know about robotics, physics, computer science, and so on. You gave a good example of that. I've seen the Empire Robotic product in action, able to grab a lot of different pieces.

It's still interesting only for big production chains, not for the dishes !