Robotics is not as cheap as you think. I've brought systems in to factories to do seemingly simple things that cost easily $250k without batting an eye. You can pay someone more than minimum wage for many years for that kind of money.
What would robots capable of stocking shelves cost? $500k? It's just not worth it. McDonalds has how many thousands of stores - and they don't bother automating that much. Amazon.com still uses people to pick their orders (for the most part). The companies aren't stupid, someone has run the numbers and figured out that robots are too expensive.
It's not even Moore's law thing. Computer are plenty powerful. You can make a robot do these things. But by the time you buy good quality motors, batteries, and sensors, it gets expensive no matter how cheap the computer is.
Industrial robots are heavy and thus expensive because they lack visual servoing. Heartland robotics plans to bring out a lightweight low cost robot this year. Also, the predator algorithm invented last year can do real time visual servoing that wasn't possible before. Add in a person to cover in the gaps and you might be golden.
If the above is true, why hasn't it been done you might ask.
I might be wrong, but you could have the said the same thing about why a search engine that didn't suck wasn't around in 1997 - why didn't microsoft, or yahoo, or ibm have that, or even buy it from sergey and larry?
What would robots capable of stocking shelves cost? $500k? It's just not worth it. McDonalds has how many thousands of stores - and they don't bother automating that much. Amazon.com still uses people to pick their orders (for the most part). The companies aren't stupid, someone has run the numbers and figured out that robots are too expensive.
It's not even Moore's law thing. Computer are plenty powerful. You can make a robot do these things. But by the time you buy good quality motors, batteries, and sensors, it gets expensive no matter how cheap the computer is.