DMCA
A POMDP Approximation Algorithm that Anticipates the Need to Observe (2000)
| Venue: | In Pacific Rim International Conference on Artificial Intelligence |
| Citations: | 1 - 0 self |
BibTeX
@TECHREPORT{Bayer00apomdp,
author = {Valentina Bayer and Thomas Dietterich},
title = {A POMDP Approximation Algorithm that Anticipates the Need to Observe},
institution = {In Pacific Rim International Conference on Artificial Intelligence},
year = {2000}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
This paper introduces the even-odd POMDP, an approximation to POMDPs in which the world is assumed to be fully observable every other time step. The even-odd POMDP can be converted into an equivalent MDP, the 2MDP, whose value function, V 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 POMDP's optimal value function that is at least as good as methods based on the optimal value function of the underlying MDP.
