Dirichlet-based Dynamic Movement Primitives for encoding periodic motions with predefined accuracy

In this work, the utilization of Dirichlet (periodic sinc) base functions in DMPs for encoding periodic motions is proposed. By utilizing such kernels, we are able to analytically compute the minimum required number of kernels based only on the predefined accuracy, which is a hyperparameter that can be intuitively selected. The computation of the minimum required number of kernels is based on the frequency content of the demonstrated motion. The learning procedure essentially consists of the sampling of the demonstrated trajectory. The approach is validated through simulations and experiments with the KUKA LWR4+ robot, which show that utilizing the automatically calculated number of basis functions, the pre-defined accuracy is achieved by the proposed DMP model.

[1]  João Silvério,et al.  Fourier movement primitives: an approach for learning rhythmic robot skills from demonstrations , 2020, Robotics: Science and Systems.

[2]  Zoe Doulgeri,et al.  Progressive Automation of Periodic Movements , 2019, HFR.

[3]  Dimitrios Papageorgiou,et al.  Sinc-Based Dynamic Movement Primitives for Encoding Point-to-point Kinematic Behaviors , 2018, 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).

[4]  Jun Morimoto,et al.  Orientation in Cartesian space dynamic movement primitives , 2014, 2014 IEEE International Conference on Robotics and Automation (ICRA).

[5]  S. Schaal,et al.  Dynamical Movement Primitives: Learning Attractor Models for Motor Behaviors , 2013, Neural Computation.

[6]  Andrej Gams,et al.  On-line periodic movement and force-profile learning for adaptation to new surfaces , 2010, 2010 10th IEEE-RAS International Conference on Humanoid Robots.

[7]  Panagiotis K. Artemiadis,et al.  Human arm impedance: Characterization and modeling in 3D space , 2010, 2010 IEEE/RSJ International Conference on Intelligent Robots and Systems.

[8]  Andrej Gams,et al.  On-line learning and modulation of periodic movements with nonlinear dynamical systems , 2009, Auton. Robots.

[9]  Jun Nakanishi,et al.  Movement imitation with nonlinear dynamical systems in humanoid robots , 2002, Proceedings 2002 IEEE International Conference on Robotics and Automation (Cat. No.02CH37292).

[10]  Asoke K. Nandi,et al.  Notes on the interpolation of discrete periodic signals using sinc function related approaches , 2000, IEEE Trans. Signal Process..

[11]  Christopher G. Atkeson,et al.  Constructive Incremental Learning from Only Local Information , 1998, Neural Computation.

[12]  Thomas Schanze,et al.  Sinc interpolation of discrete periodic signals , 1995, IEEE Trans. Signal Process..

[13]  Tadej Petric,et al.  Teaching robots to cooperate with humans in dynamic manipulation tasks based on multi-modal human-in-the-loop approach , 2014, Auton. Robots.

[14]  S. Schaal.  Dynamic Movement Primitives -A Framework for Motor Control in Humans and Humanoid Robotics , 2006 .

[15]  C.E. Shannon,et al.  Communication in the Presence of Noise , 1949, Proceedings of the IRE.