A POMDP approximation algorithm that anticipates the need to observe

Technical Report

Público Deposited

Citeable URL: https://ir.library.oregonstate.edu/concern/technical_reports/gm80hw69f

Descriptions

Attribute Name	Values
Creator	Oregon State University. Dept. of Computer Science Zubek, Valentina Bayer Dietterich, Thomas Glen
Abstract	This paper introduces the even-odd POMDP an approximation to POMDPs Partially Observable Markov Decision Problems in which the world is assumed to be fully observable every other time step. This approximation works well for problems with a delayed need to observe. The even-odd POMDP can be converted into an equivalent MDP the 2MDP whose value function, V*[subscript 2MDP], can be combined online with a 2-step lookahead search to provide a good POMDP policy. We prove that this gives an approximation to the POMDPs optimal value function that is at least as good as methods based on the optimal value function of the underlying MDP. We present experimental evidence that the method finds a good policy for a POMDP with states and observations. Keywords: Partially Observable Markov Decision Problems, Even-odd POMDP, POMDP
Resource Type	Research Paper
Fecha Disponible	2012-05-30T22:39:04+00:00
Fecha de Emisión	2004-07-05
Series	Technical report (Oregon State University. Department of Computer Science)
Subject	Markov processes Approximation algorithms
Declaración de derechos	Copyright Not Evaluated
Funding Statement (additional comments about funding)	This research was supported by AFOSR F49620-9810375.
Publisher	Corvallis, OR : Oregon State University, Dept. of Computer Science
Peer Reviewed	No
Language	English [eng]
Replaces	http://hdl.handle.net/1957/29461

Miniatura	Título	Fecha de subida	Visibilidad	Acciones
	A_POMDP_approximation_algorithm_that_anticipates_the_need_to_observe_2004.pdf	2017-07-18	Público	Descargar