Incremental Random-Features-Based Inverse Dynamics Learning on iCub

Описание: iRRLS is an inverse dynamics estimation application for the iCub humanoid robot, based on the YARP, iCub-main and GURLS libraries.

This tutorial shows the online operation of the application on the dataset gathered from the right arm of the robot while performing a randomly generated motion.

For more information, see:

[1] GURLS repo: https://github.com/LCSL/GURLS
[2] iRRLS repo: https://github.com/Ommac/iRRLS
[3] http://www.iit.it/images/images/icub-...
[4] Gijsberts et al., ‘‘Incremental Learning of Robot Dynamics using Random Features’’, ICRA 2011
[5] Duy Nguyen-Tuong, Jan Peters, “Using Model Knowledge for Learning Inverse Dynamics”, ICRA 2010