Iterative Motion Primitive Learning and Refinement by Compliant Motion Control

We present an approach for motion primitive learning and refinement for a humanoid robot. The proposed teaching method starts with observational learning and applies iterative kinesthetic motion refinement. Observational learning is realized by a marker control approach. Kinesthetic teaching is supported by introducing the motion refinement tube in order to avoid to disturb other joints accidentally. The proposed teaching method allows for iterative execution and motion refinement using a forgetting factor. On the realtime control level, the kinesthetic teaching is handled by a customized Cartesian impedance controller, which combines tracking performance with compliant physical interaction and allows to implement soft boundaries for the motion refinement. A new method for continuous generation of motions from a hidden Markov model (HMM) representation of motion primitives is proposed. The proposed methods were implemented and tested using DLR’s humanoid upper-body robot Justin.