RT Conference Proceedings
T1 Programming-by-demonstration and adaptation of robot skills by fuzzy-time-modeling
A1 Palm, Rainer
A1 Iliev, Boyko
AB Complex robot tasks can be partitioned into motionprimitives or robot skills that can directly be learned andrecognized through Programming-by-Demonstration (PbD) by ahuman operator who demonstrates a set of reference skills. Robotmotions are recorded by a data-capturing system and modeledby a specific fuzzy clustering and modeling technique whereskill models use time instants as inputs and operator actionsas outputs. In the recognition phase the robot identifies the skillshown by the operator in a novel test demonstration.Skill models are updated online during the execution of skillsusing the Broyden update formula. This method is extended forfuzzy models especially for time cluster models. The updatedmodel is used for further executions of the same skill.
PB IEEE
SN 978-1-4244-9884-0
YR 2011
FD 2011-07
LK https://hdl.handle.net/10016/13982
UL https://hdl.handle.net/10016/13982
LA eng
NO Proceedings of: 2011 IEEE Workshop on Robotic Intelligence in Informationally Structured Space (RiiS 2011 MDCM), April 11-15, 2011, Paris (France)
NO European Community's Seventh Framework Program
DS e-Archivo
RD 18 may. 2024