Generalized Task-Parameterized Movement Primitives

Programming by demonstrations has recently gained much attention due to its user-friendly and natural way to transfer human skills to robots. In order to facilitate the learning of multiple demonstrations and meanwhile generalize to new situations, a task-parameterized movement learning has been recently developed [4], which has achieved reliable performance in areas such as human-robot collaboration and robot bimanual operation. However, the crucial task frames and associated task parameters in this learning framework are often set based on human experience, which renders three problems that have not been addressed yet: (i) task frames are treated equally without considering the task priorities; (ii) task parameters are defined without considering additional task constraints, e.g., robot joint limits and motion smoothness; (iii) a fixed number of task frames are pre-defined regardless some of them are redundant or even irrelevant for the task at hand. In this paper, we generalize the task-parameterized learning by addressing the aforementioned problems. Moreover, we provide an alternative way to refine and adapt previously learned robot skills, which allows us to work on a low dimensional space. Several examples are studied in simulated and real robotic systems, showing the applicability of our approach.

[1]  Stefan Schaal,et al.  Learning variable impedance control , 2011, Int. J. Robotics Res..

[2]  Jochen J. Steil,et al.  Task-level imitation learning using variance-based movement optimization , 2009, 2009 IEEE International Conference on Robotics and Automation.

[3]  Aude Billard,et al.  Reinforcement learning for imitating constrained reaching movements , 2007, Adv. Robotics.

[4]  Darwin G. Caldwell,et al.  Learning bimanual end-effector poses from demonstrations using task-parameterized dynamical systems , 2015, 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).

[5]  Carme Torras,et al.  Learning Physical Collaborative Robot Behaviors From Human Demonstrations , 2016, IEEE Transactions on Robotics.

[6]  Marc Toussaint,et al.  Optimization of sequential attractor-based movement for compact behaviour generation , 2007, 2007 7th IEEE-RAS International Conference on Humanoid Robots.

[7]  Stefan Schaal,et al.  A Generalized Path Integral Control Approach to Reinforcement Learning , 2010, J. Mach. Learn. Res..

[8]  Andrew W. Moore,et al.  Locally Weighted Learning , 1997, Artificial Intelligence Review.

[9]  Jun Nakanishi,et al.  Dynamical Movement Primitives: Learning Attractor Models for Motor Behaviors , 2013, Neural Computation.

[10]  Jan Peters,et al.  Probabilistic Movement Primitives , 2013, NIPS.

[11]  Nikolaos G. Tsagarakis,et al.  COMpliant huMANoid COMAN: Optimal joint stiffness tuning for modal frequency control , 2013, 2013 IEEE International Conference on Robotics and Automation.

[12]  Sylvain Calinon,et al.  A tutorial on task-parameterized movement learning and retrieval , 2015, Intelligent Service Robotics.

[13]  David A. Cohn,et al.  Active Learning with Statistical Models , 1996, NIPS.

[14]  Brett Browning,et al.  A survey of robot learning from demonstration , 2009, Robotics Auton. Syst..

[15]  Olivier Sigaud,et al.  Robot Skill Learning: From Reinforcement Learning to Evolution Strategies , 2013, Paladyn J. Behav. Robotics.