Reinforcement Learning Control with TimeDependent Agent Dynamics Chapter

abstract

• Reinforcement learning-based control systems have the advantage of being modelfree. When used for control of dynamical systems, approximations of the time dynamics can be used to improve the policy development. This chapter proposes and develops a Q-Learning-based method for learning unknown information about the time scale properties of a dynamical system. Two candidate algorithm are proposed. The first determines optimal sample times for sampled data systems, and is demonstrated with a simulation of an inverted pendulum cart. The second algorithm learns a first-order approximation for a dynamical system, and uses the approximation to improve the control of a heterogenous multiagent system. This algorithm is demonstrated with a conceptual experiment involving such a system. 2013 The Institute of Electrical and Electronics Engineers, Inc.

authors

• Valasek, John

author list (cited authors)

• Kirkpatrick, K., & Valasek, J.

citation count

• 1

complete list of authors

• Kirkpatrick, Kenton||Valasek, John

Book Title

• Reinforcement Learning and Approximate Dynamic Programming for Feedback Control

publication date

• December 2012

publisher

• Wiley  Publisher

keywords

• Artificial Intelligence
• Control
• Dynamics
• Machine Learning
• Multiagent
• Q-learning
• Reinforcement Learning
• Sampled-data Systems
• Sampling
• System Identification

Digital Object Identifier (DOI)

• 10.1002/9781118453988.ch10

International Standard Book Number (ISBN) 13

• 9781118104200

URI

• https://hdl.handle.net/1969.1/149493

start page

• 203

end page

• 220

URL

• http://dx.doi.org/10.1002/9781118453988.ch10