Reinforcement Learningfor HumanoidRobotics

Jan Peters, Sethu Vijayakumar, Stefan Schaal

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Reinforcement learning of fersoneof the most general frame-work to take traditional robotics towards true autonomy and versatility.However, applying reinforcement learning to high dimensional movement systems like humanoid robots remains an unsolved problem. In this paper, we discuss different approaches of reinforcement learning in terms of their applicability in humanoid robotics. Methods can be coarsely classified into three different categories, i.e., greedy methods, ‘vanilla’ policy gradient methods, and natural gradient methods. We discuss that greedy methods are not likely to scale into the domain humanoid robotics as they are problematic when used with function approximation. ‘Vanilla’ policy gradient methods on the other hand have been successfully applied on real-world robots including at least one humanoid robot [3]. We demonstrate that these methods can be significantly improved using the natural policy gradient instead of the regular policy gradient. A derivation of the natural policy gradient is provided, proving that the average policy gradient of Kakade [10] is indeed the true natural gradient. A general algorithm for estimating the natural gradient, the Natural Actor-Critic algorithm, is introduced. This algorithm converges to the nearest local minimum of the cost function with respect to the Fisher information metric under suitable conditions. The algorithm out performs non-natural policy gradients by far in a cart-pole balancing evaluation, and for learning nonlinear dynamic motor primitives for humanoid robot control. It offers a promising route for the development of reinforcement learning for truly high-dimensionally continuous state-action systems.
Original languageEnglish
Title of host publicationHumanoids2003, Third IEEE-RAS International Conference on Humanoid Robots
Number of pages20
Publication statusPublished - 29 Sept 2003
Event3rd IEEE-RAS InternationalConference on Humanoid Robots - Karlsruhe, Germany
Duration: 29 Sept 200330 Sept 2003

Conference

Conference3rd IEEE-RAS InternationalConference on Humanoid Robots
Abbreviated titleHumanoids 2003
Country/TerritoryGermany
CityKarlsruhe
Period29/09/0330/09/03

Fingerprint

Dive into the research topics of 'Reinforcement Learningfor HumanoidRobotics'. Together they form a unique fingerprint.

Cite this