Reinforcement Learning for Motor Primitives

Abstract

Humans demonstrate a large variety of very complicated motor skills in their day-to-day life. Their agility and adaptability to new control task remains unmatched by the millions of robots laboring on factory floors and roaming research labs. Achieving the abilities of learning and improving new motor skills has become an essential component in order to get a step closer to human-like motor skills. If future robots could acquire their basic task by imitating human demonstrations and subsequently self-improve by trial and error, such robot learning would result into more interesting robot applications as well as large productivity gains in industry. Recent progress in the area of machine learning has yielded several important tools for making progress towards this vision for the future. Two of these recent developments are a novel framework for representing motor primitives using dynamical systems presented in [Ijspeert et al., 2002a,b, 2003, Schaal et al., 2004] and the reduction of reward-related self-improvement to reward-weighted regression by Peters and Schaal [2006c, 2007b]. To date, these two important methods have only been used separately in robot learning; when using them in combination, this could yield a strong basis for learning motor skills. In this