Publikation
Automated Robot Skill Learning from Demonstration for Various Robot Systems
Lisa Gutzeit; Alexander Fabisch; Christoph Petzold; Hendrik Wiese; Frank Kirchner
In: KI 2019: Advances in Artificial Intelligence. German Conference on Artificial Intelligence (KI-2019), September 23-26, Kassel, Germany, Pages 168-181, LNAI, Springer, 2019.
Zusammenfassung
Transferring human movements to robotic systems is of high interest
to equip the systems with new behaviors without expert knowledge. Typ-
ically, skills are often only learned for a very specific setup and a certain
robot. We propose a modular framework to learn skills that is applica-
ble on different robotic systems without adaptations. Our work builds
on the recently introduced BesMan Learning Platform, which comprises
the full workflow to transfer human demonstrations to a system, includ-
ing automatized behavior segmentation, imitation learning, reinforcement
learning for motion refinement, and methods to generalize to related tasks.
For this paper, we extend this approach in order that different skills can
be imitated by various systems in an automated fashion with a minimal
amount of configuration, e.g., definition of the target system and environ-
ment. For this, we focus on the imitation of the demonstrated movements
and show their transferability without movement refinement. We demon-
strate the generality of the approach on a large dataset, consisting of about
700 throwing demonstrations. Nearly all of these human demonstrations
are successfully transferred to four different robot target systems, namely
Universal Robot’s UR5 and UR10, KUKA LBR iiwa, and DFKI’s robot
COMPI. An analysis of the quality of the imitated movement on the real
UR5 robot shows that useful throws can be executed on the system which
can be used as starting points for further movement refinement.