Learning for Control from Multiple Demonstrations
Published on Aug 12, 200810522 Views
We consider the problem of learning to follow a desired trajectory when given a small number of demonstrations from a sub-optimal expert. We present an algorithm that (i) extracts the---initially unkn