Automated Derivation of Primitives for Movement Classification

Abstract

1 Introduction Programming and interacting with robots, especially humanoid ones, is a complex problem. Using learning toaddress this problem is a popular approach, but the high dimensionality of humanoid control makes the approach prohibitively slow. Imitation, the process of learning new movement patterns and skills by observation, is a promis-ing alternative. The ability to imitate enables a robot to greatly reduce the space of possible trajectories to a subset that approximates that of the observed demonstration. Refinement through trial and error is still likely to be re-quired, but in a greatly reduced learning space.