Improving Reinforcement Learning Speed for Robot Control
- 1 October 2006
- conference paper
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 3172-3177
- https://doi.org/10.1109/iros.2006.282341
Abstract
Reinforcement learning (R-L) is an intuitive way of programming well-suited for use on autonomous robots because it does not need to specify how the task has to be achieved. However, RL remains difficult to implement in realistic domains because of its slowness in convergence. In this paper, we develop a theoretical study of the influence of some RL parameters over the learning speed. We also provide experimental justifications for choosing the reward function and initial Q-values in order to improve RL speed within the context of a goal-directed robot taskKeywords
This publication has 6 references indexed in Scilit:
- Reinforcement learning for motion control of humanoid robotsPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2005
- Multi-agent quadrotor testbed control design: integral sliding mode vs. reinforcement learningPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2005
- Effective reinforcement learning for mobile robotsPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2003
- On amount and quality of bias in reinforcement learningPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2003
- Reinforcement Learning in Continuous Time and SpaceNeural Computation, 2000
- Technical Note: Q-LearningMachine Learning, 1992