You're talking about different things.

Computer vision to detect red and green blocks in a controlled environment is completely trivial.

Driving a robot with human hands to pick up and manipulate blocks is still advanced research.