Real-Time Local GP Model Learning

For many applications in robotics, accurate dynamics models are essential. However, in some applications, e.g., in model-based tracking control, precise dynamics models cannot be obtained analytically for sufficiently complex robot systems. In such cases, machine learning offers a promising alternative for approximating the robot dynamics using measured data. However, standard regression methods such as Gaussian process regression (GPR) suffer from high computational complexity which prevents their usage for large numbers of samples or online learning to date. In this paper, we propose an approximation to the standard GPR using local Gaussian processes models inspired by [Vijayakumar et al(2005)Vijayakumar, D’Souza, and Schaal, Snelson and Ghahramani(2007)]. Due to reduced computational cost, local Gaussian processes (LGP) can be applied for larger sample-sizes and online learning. Comparisons with other nonparametric regressions, e.g., standard GPR, support vector regression (SVR) and locally weighted projection regression (LWPR), show that LGP has high approximation accuracy while being sufficiently fast for real-time online learning.

[1]  Francis L. Merat,et al.  Introduction to robotics: Mechanics and control , 1987, IEEE J. Robotics Autom..

[2]  Anthony Widjaja,et al.  Learning with Kernels: Support Vector Machines, Regularization, Optimization, and Beyond , 2003, IEEE Transactions on Neural Networks.

[3]  John J. Craig,et al.  Introduction to robotics - mechanics and control (2. ed.) , 1989 .

[4]  Mark W. Spong,et al.  Robot dynamics and control , 1989 .

[5]  Jun Nakanishi,et al.  Composite adaptive control with locally weighted statistical learning , 2005, Neural Networks.

[6]  Zoubin Ghahramani,et al.  Local and global sparse Gaussian process approximations , 2007, AISTATS.

[7]  E. Ziegel,et al.  Artificial intelligence and statistics , 1986 .

[8]  Matthias W. Seeger,et al.  Bayesian Gaussian process models : PAC-Bayesian generalisation error bounds and sparse approximations , 2003 .

[9]  Olivier Sigaud,et al.  From Motor Learning to Interaction Learning in Robots , 2010, From Motor Learning to Interaction Learning in Robots.

[10]  Jun Zhang,et al.  Motor Learning at Intermediate Reynolds Number: Experiments with Policy Gradient on the Flapping Flight of a Rigid Wing , 2010, From Motor Learning to Interaction Learning in Robots.

[11]  Stefan Schaal,et al.  Incremental Online Learning in High Dimensions , 2005, Neural Computation.

[12]  Stefan Schaal,et al.  Real-time robot learning with locally weighted statistical learning , 2000, Proceedings 2000 ICRA. Millennium Conference. IEEE International Conference on Robotics and Automation. Symposia Proceedings (Cat. No.00CH37065).

[13]  Chih-Jen Lin,et al.  LIBSVM: A library for support vector machines , 2011, TIST.

[14]  Lehel Csató,et al.  Sparse On-Line Gaussian Processes , 2002, Neural Computation.

[15]  M. Opper.  Sparse Online Gaussian Processes , 2008 .

[16]  Carl E. Rasmussen,et al.  Gaussian processes for machine learning , 2005, Adaptive computation and machine learning.

[17]  Giulio Sandini,et al.  Learning to Exploit Proximal Force Sensing: A Comparison Approach , 2010, From Motor Learning to Interaction Learning in Robots.

[18]  Duy Nguyen-Tuong,et al.  Computed torque control with nonparametric regression models , 2008, 2008 American Control Conference.

[19]  John W. Roberts,et al.  Motor Learning at Intermediate Reynolds Number: Experiments with Policy Gradient on the Flapping Flight of a Rigid Wing , 2010, From Motor Learning to Interaction Learning in Robots.

[20]  Stefan Schaal,et al.  Scalable Techniques from Nonparametric Statistics for Real Time Robot Learning , 2002, Applied Intelligence.