Controlling a robot during interaction with a human

This system teaches a robot to perform tasks by controlling its movements around objects held in place by stationary fixtures. It uses a computer brain, called a neural network, which learns how to move and grasp objects safely. The learning process specifically involves planning out collision-free movements and pre-defined ways to grab things, ensuring the robot avoids hitting the stationary holders.