SkyAI: Highly Modularized Reinforcement Learning Library
SkyAI is an open source software library of reinforcement learning methods. The distinct feature is its modular architecture which realizes high execution-speed enough for real robot systems and high flexibility to design learning systems. In addition, SkyAI is designed so that the modular structure can be changed during execution.
Currently, the development version of SkyAI is released. You can download it from SourceForge. Please visit the website of SkyAI for the detail.
Example of six tasks learned using SkyAI:
Related Papers
- Akihiko Yamaguchi, and Tsukasa Ogasawara:
A Novel Robot-Task-Description for a Variety of Dynamic Behaviors,
in Proceedings of the 2012 IEEE/SICE International Symposium on System Integration (SII2012), pp.86-92, Fukuoka, Japan, December 16, 2012. [final-draft] - Akihiko Yamaguchi, and Tsukasa Ogasawara:
SkyAI: Highly Modularized Reinforcement Learning Library ---Concepts, Requirements, and Implementation---,
in Proceedings of the 10th IEEE-RAS International Conference on Humanoid Robots (Humanoids2010), pp.118-123, Nashville, TN, US, 2010. [final-draft] - Akihiko Yamaguchi, Tsukasa Ogasawara:
SkyAI: Highly Modularized Reinforcement Learning Library ---Script Interface---,
in Proceedings of the 28th Annual Conference of the Robotics Society of Japan, RSJ2010AC1I1-4, Nagoya, September, 2010. (in Japanese) [PDF]