Learning robot dynamics with Kinematic Bézier Maps

The previously presented Kinematic Bézier Maps (KBM) are a machine learning algorithm that has been tailored to efficiently learn the kinematics of redundant robots. This algorithm relies upon a representation based on projective geometry that uses a special set of polynomial functions borrowed from the field of Computer Aided Geometric Design (CAGD). So far, it has only been possible to learn a model of the forward kinematics function. In this paper, we show how the KBM algorithm can be modified to learn the robot's equation of motion and, hence, its inverse dynamic model. Results from experiments with a simulated serial robot manipulator are presented that clearly show the advantages of our approach compared to general function approximation methods.

[1]  Olivier Sigaud,et al.  On-line regression algorithms for learning mechanical models of robots: A survey , 2011, Robotics Auton. Syst..

[2]  Stefan Klanke,et al.  PSOM+ : Parametrized Self-Organizing Maps for noisy and incomplete data , 2005 .

[3]  Aude Billard,et al.  On Learning, Representing, and Generalizing a Task in a Humanoid Robot , 2007, IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics).

[4]  Bernhard Schölkopf,et al.  Learning Inverse Dynamics: a Comparison , 2008, ESANN.

[5]  Alejandro Hernández Arieta,et al.  Body Schema in Robotics: A Review , 2010, IEEE Transactions on Autonomous Mental Development.

[6]  Dana Kulic,et al.  Learning inverse dynamics for redundant manipulator control , 2010, 2010 International Conference on Autonomous and Intelligent Systems, AIS 2010.

[7]  Stefan Ulbrich,et al.  Kinematic Bézier Maps , 2012, IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics).

[8]  Stefan Ulbrich,et al.  General Robot Kinematics Decomposition Without Intermediate Markers , 2012, IEEE Transactions on Neural Networks and Learning Systems.

[9]  Stefan Schaal,et al.  Learning inverse kinematics , 2001, Proceedings 2001 IEEE/RSJ International Conference on Intelligent Robots and Systems. Expanding the Societal Role of Robotics in the the Next Millennium (Cat. No.01CH37180).

[10]  Jan Peters,et al.  Local Gaussian process regression for real-time model-based robot control , 2008, 2008 IEEE/RSJ International Conference on Intelligent Robots and Systems.

[11]  John M. Hollerbach,et al.  A Recursive Lagrangian Formulation of Maniputator Dynamics and a Comparative Study of Dynamics Formulation Complexity , 1980, IEEE Transactions on Systems, Man, and Cybernetics.

[12]  Helge J. Ritter,et al.  Rapid learning with parametrized self-organizing maps , 1996, Neurocomputing.

[13]  J. Eisert,et al.  Extracting dynamical equations from experimental data is NP hard. , 2010, Physical review letters.

[14]  Aristidis Likas,et al.  Unsupervised Learning of Gaussian Mixtures Based on Variational Component Splitting , 2007, IEEE Transactions on Neural Networks.

[15]  Mark Whitty,et al.  Robotics, Vision and Control. Fundamental Algorithms in MATLAB , 2012 .

[16]  Jan Peters,et al.  Model learning for robot control: a survey , 2011, Cognitive Processing.

[17]  Stefan Ulbrich,et al.  Rapid learning of humanoid body schemas with Kinematic Bézier Maps , 2009, 2009 9th IEEE-RAS International Conference on Humanoid Robots.