Programming-by-demonstration and adaptation of robot skills by fuzzy-time-modeling

Abstract

Complex robot tasks can be partitioned into motion primitives or robot skills that can directly be learned and recognized through Programming-by-Demonstration (PbD) by a human operator who demonstrates a set of reference skills. Robot motions are recorded by a data-capturing system and modeled by a specific fuzzy clustering and modeling technique where skill models use time instants as inputs and operator actions as outputs. In the recognition phase the robot identifies the skill shown by the operator in a novel test demonstration. Skill models are updated online during the execution of skills using the Broyden update formula. This method is extended for fuzzy models especially for time cluster models. The updated model is used for further executions of the same skill.