POMCP-based decentralized spatial task allocation algorithms for partially observable environments.
Saved in:
| Title: | POMCP-based decentralized spatial task allocation algorithms for partially observable environments. |
|---|---|
| Authors: | Amini, Sara, Palhang, Maziar, Mozayani, Nasser |
| Source: | Applied Intelligence; May2023, Vol. 53 Issue 10, p12613-12631, 19p |
| Subject Terms: | ALGORITHMS, DISTRIBUTED algorithms |
| Abstract: | Spatial task allocation has many applications in realistic multi-robot systems and has been studied for several years by many researchers. However, most of the researches conducted so far focused on centralized algorithms or fully observable environments. This paper proposes a decentralized task allocation algorithm for partially observable worlds that extends Partially Observable Monte Carlo Planning (POMCP) to a multi-agent world using a self-absorbed view. This extension is performed in two different ways: Using a POMCP tree with single actions but encoding the other agents' actions implicitly in the tree edges, which we call SPOMCP. The second view, called GPOMCP, is to apply a POMCP tree with action profiles besides doing a coordinate-wise optimization to find the best single action. Experimental results show that SPOMCP and GPOMCP do fairly well compared to other algorithms. SPOMCP consumes more time than GPOMCP for each iteration since it needs to take time to infer other robots' actions. GPOMCP does not do such an inference step because its search tree is built by action profiles. Comparison with a centralized POMCP (CPOMCP) indicates that while both SPOMCP and GPOMCP are fully decentralized, they can reach the performance of CPOMCP. However, SPOMCP and GPOMCP consume much less time than CPOMCP. [ABSTRACT FROM AUTHOR] |
| Copyright of Applied Intelligence is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Complementary Index |
Full Text Finder 
Nájsť tento článok vo Web of Science Be the first to leave a comment!