A Framework for Reinforcement Learning on Real Robots

Learning on real robots in an real, unaltered environment provides an extremely challenging problem. Many of the simplifying assumptions made in other areas of learning cannot be applied, the environment is often unforgiving and progress must be made in real-time. In this paper, we introduce a framework for learning on real-world robots that addresses these issues and allows some standard learning techniques to be used more easily on robots. We also present and discuss some preliminary results obtained with this framework.

• Type: Conference
• Year: 1998
• Where: AAAI
• Authors: William D. Smart, Leslie Pack Kaelbling
• Keywords: AAAI 1998, Challenging Problem, Intelligent Agents, Real Robots, Unaltered Environment