H. Geffner and B. Bonet, "Solving large POMDPs using real time dynamic programming," in Working Notes Fall AAAI Symposium on POMDPs, pp. 61--68, 1998. Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
A Survey of POMDP Solution Techniques - Murphy (2000) (1 citation) (Correct)
.... the values of the leaves as 0 (due to discounting) An alternative is to use an admissible heuristic (lower bound) for the value of the leaves (e.g. V MDP (b) and to update the values of the intermediate nodes as in Real Time Dynamic Programming [BBS95] This is the approach taken in [GB98], who discretizes b, so V can be represented as a table. Unfortunately, receeding horizon control only works for discrete (or discretized) actions (except for special cases, like linear quadratic Gaussian) 4 Direct policy search The most widely used approach to solving MDPs is an indirect one, ....
H. Ge ner and B. Bonet. Solving large POMDPs using real time dynamic programming. In Fall AAAI Symp. on POMDPs, 1998.
Planning and Control in Artificial Intelligence: A Unifying.. - Bonet, Geffner (2001) Self-citation (Bonet Geffner) (Correct)
.... (MDPs) and Partially Observable MDPs (POMDPs) borrows from the work in Dynamic Programming [51, 6] Finally, the use heuristic search algorithms has long tradition in AI (e.g. 46] even though the use of such algorithms for solving Strips planning problems, MDPs, and POMDPs is more recent [3, 13, 11, 32]. In this paper we provide a coherent integration of these various elements and argue that the result provides a natural framework for modeling and solving planning problems under a variety of conditions, including deterministic and probabilistic actions, and complete, partial or null sensing. We ....
.... S, is infinite and continuous (unlike the finite state mdps we have considered so far) The known optimal pomdp algorithms can thus solve very small problems only (e.g. see [17] Non optimal methods, on the other hand, scale up better and often produce reasonable results for non trivial problems [33, 31, 59, 11]. Here we present an algorithm that is the adaptation of rtdp for solving belief mdps. We call such algorithm rtdp bel and show it in Fig. 6. Details on rtdp bel can be found in [11, 9, 10] where the algorithm is used to solve a variety of problems: planning problems with sensing, robot ....
[Article contains additional citation context not shown here]
B. Bonet and H. Geffner. Solving large POMDPs using real time dynamic programming. In Proc. AAAI Fall Symp. on POMDPs, 1998. 24
Planning with Incomplete Information as Heuristic Search in.. - Bonet, Geffner (2000) (31 citations) Self-citation (Bonet Geffner) (Correct)
No context found.
Bonet, B., and Geffner, H. 1998b. Solving large POMDPs using real time dynamic programming. In Proc. AAAI Fall Symp. on POMDPs.
Learning sorting and decision trees with POMDPs - Bonet, Geffner (1998) (2 citations) Self-citation (Geffner Bonet) (Correct)
No context found.
Geffner, H., and Bonet, B. 1998b. Solving large POMDPs using real time dynamic programming.
Algorithms for Partially Observable Markov Decision Processes - Zhang (2001) (Correct)
No context found.
H. Geffner and B. Bonet, "Solving large POMDPs using real time dynamic programming," in Working Notes Fall AAAI Symposium on POMDPs, pp. 61--68, 1998.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC