Path Learning by Demonstration for Iterative Human-Robot Interaction with Uncertain Time Durations

This paper presents a path learning method through physical human-robot interaction (pHRI) based on stretch-compression iterative control (ILC) scheme and contouring impedance control. The robot learns task desired by the human user kinaesthetic interface provides assistance to in repetitive interactions. Due uncertainty of user’s force motion, time duration each iteration may be different, so novel ILC stretch compression operation is proposed update reference trajectory robotic manipulator. By attaching Frenet-Serret frame point path, decomposed into tangential direction position normal or binormal constraining path. Experiments 7-DOF Sawyer are carried out show effectiveness robustness method.