I worked for a company that did automated warehouses once. Their development over many years went something like this:
Fully manual: people would pick stuff from shelves and put it in baskets. It was organized in a complicated way but that's not very important, it was manual in the end.
Mixed: they had packing stations. Worker would stand in front of a screen and plastic create would come on conveyor belts and stop in front of them. They would have plastic bags below the basket. Instructions on the screen would tell them what to pick up. For example a crate would come full of soda cans and they would see "put 3 coca-cola cans in the bag" in front of them. The bagging process is very hard to automate because robots have trouble recognizing and grabbing things. The crate delivery system was fully automated and very complex. It would take up to 20 minutes to take a crate from the warehouse and deliver it to packing station so everything had to be synchronized so that all the crates needed to fulfil and order would come to specific packing station one after another. The move from manual to mixed models cost them hundredths of millions to develop. They had to build new warehouse from scratch. The mixed model still had lots of people dealing with edge cases like cutting cheese or handling fish.
Another mixed: they had this huge cube like structure with small elevators moving plastic crates up and down inside of it and small robots moving the crates between stacks on top of it. You could tell it which create you needed and the cube would pick it up and deliver. It was the same as the huge warehouse as in it would deliver the crates in specific order but was a lot smaller. People would still have to bag it manually. Again, this was build in a new warehouse from scratch.
So as you see the thing is moving from one model to another is really complicated and requires rebuilding everything. They have tons of warehouses optimized for people so it makes more sense for them to build humanoid robots than rebuild all the warehouses.
I worked for a company that did automated warehouses once. Their development over many years went something like this:
So as you see the thing is moving from one model to another is really complicated and requires rebuilding everything. They have tons of warehouses optimized for people so it makes more sense for them to build humanoid robots than rebuild all the warehouses.
Thanks for the insight!