Approximate Solutions for Factored Dec-POMDPs with Many Agents
| Authors | |
|---|---|
| Publication date | 2013 |
| Journal | BNAIC |
| Event | 25th Belgium-Netherlands Artificial Intelligence Conference (BNAIC 2013) |
| Volume | Issue number | 25 |
| Pages (from-to) | 340-341 |
| Organisations |
|
| Abstract | Dec-POMDPs are a powerful framework for planning in multiagent systems, but are provably intractable to solve. This paper proposes a factored forward-sweep policy computation method that tackles the stages of the problem one by one, exploiting weakly coupled structure at each of these stages. An empirical evaluation shows that the loss in solution quality due to these approximations is small and that the proposed method achieves unprecedented scalability, solving Dec-POMDPs with hundreds of agents. |
| Document type | Article |
| Note | Extended Abstract Proceedings title: Proceedings of the 25th Benelux conference on Artificial Intelligence Publisher: Delft University of Technology Place of publication: Delft Editors: K. Hindriks, M. de Weerdt, B. van Riemsdijk, M. Warnier |
| Language | Dutch |
| Published at | http://bnaic2013.tudelft.nl/proceedings/papers/paper_31.pdf |
| Downloads |
paper_31
(Final published version)
|
| Permalink to this page | |