Legs that can walk: Embodiment-Based Modular Reinforcement Learning applied

Jacob, D., Polani, D. and Nehaniv, C.L. (2005) Legs that can walk: Embodiment-Based Modular Reinforcement Learning applied. In: Procs 2005 IEEE Int Symposium on Computational Intelligence in Robotics and Automation : CIRA 2005. Institute of Electrical and Electronics Engineers (IEEE), FIN, pp. 365-372. ISBN 0-7803-9355-4
Copy

Experiments to illustrate a novel methodology for reinforcement learning in embodied physical agents are described. A simulated legged robot is decomposed into structurebased modules following the authors' EMBER principles of local sensing, action and learning. The legs are individually trained to 'walk' in isolation, and re-attached to the robot; walking is then sufficiently stable that learning in situ can continue. The experiments demonstrate the benefits of the modular decomposition: state-space factorisation leads to faster learning, in this case to the extent that an otherwise intractable problem becomes learnable.


picture_as_pdf
101964.pdf
subject
Published Version

View Download

Atom BibTeX OpenURL ContextObject in Span OpenURL ContextObject Dublin Core MPEG-21 DIDL Data Cite XML EndNote HTML Citation METS MODS RIOXX2 XML Reference Manager Refer ASCII Citation
Export

Downloads