A Finite-State Controller Based Offline Solver for Deterministic POMDPs
Abstract
Deterministic partially observable Markov decision processes (DetPOMDPs) often arise in planning problems where the agent is uncertain about its environmental state but can act and observe deterministically. In this paper, we propose DetMCVI, an adaptation of the Monte Carlo Value Iteration (MCVI) algorithm for DetPOMDPs, which builds policies in the form of finite-state controllers (FSCs). DetMCVI solves large problems with a high success rate, outperforming existing baselines for DetPOMDPs. We also verify the performance of the algorithm in a real-world mobile robot forest mapping scenario.
Cite
Text
Schutz et al. "A Finite-State Controller Based Offline Solver for Deterministic POMDPs." International Joint Conference on Artificial Intelligence, 2025. doi:10.24963/IJCAI.2025/958Markdown
[Schutz et al. "A Finite-State Controller Based Offline Solver for Deterministic POMDPs." International Joint Conference on Artificial Intelligence, 2025.](https://mlanthology.org/ijcai/2025/schutz2025ijcai-finite/) doi:10.24963/IJCAI.2025/958BibTeX
@inproceedings{schutz2025ijcai-finite,
title = {{A Finite-State Controller Based Offline Solver for Deterministic POMDPs}},
author = {Schutz, Alex and You, Yang and Mattamala, Matías and Caliskanelli, Ipek and Lacerda, Bruno and Hawes, Nick},
booktitle = {International Joint Conference on Artificial Intelligence},
year = {2025},
pages = {8617-8625},
doi = {10.24963/IJCAI.2025/958},
url = {https://mlanthology.org/ijcai/2025/schutz2025ijcai-finite/}
}