[HTML][HTML] Risk-aware shielding of partially observable monte carlo planning policies

G Mazzi, A Castellini, A Farinelli - Artificial Intelligence, 2023 - Elsevier
Abstract Partially Observable Monte Carlo Planning (POMCP) is a powerful online algorithm
that can generate approximate policies for large Partially Observable Markov Decision …

Partially Observable Monte Carlo Planning with state variable constraints for mobile robot navigation

A Castellini, E Marchesini, A Farinelli - Engineering Applications of Artificial …, 2021 - Elsevier
Autonomous mobile robots employed in industrial applications often operate in complex and
uncertain environments. In this paper we propose an approach based on an extension of …

Centralizing state-values in dueling networks for multi-robot reinforcement learning mapless navigation

E Marchesini, A Farinelli - 2021 IEEE/RSJ International …, 2021 - ieeexplore.ieee.org
We study the problem of multi-robot mapless navigation in the popular Centralized Training
and Decentralized Execution (CTDE) paradigm. This problem is challenging when each …

Rule-based shielding for partially observable Monte-Carlo planning

G Mazzi, A Castellini, A Farinelli - Proceedings of the international …, 2021 - ojs.aaai.org
Abstract Partially Observable Monte-Carlo Planning (POMCP) is a powerful online algorithm
able to generate approximate policies for large Partially Observable Markov Decision …

Learning state-variable relationships for improving POMCP performance

M Zuccotto, A Castellini, A Farinelli - Proceedings of the 37th ACM …, 2022 - dl.acm.org
We address the problem of learning state-variable relationships across different episodes in
Partially Observable Markov Decision Processes (POMDPs) to improve planning …

Identification of unexpected decisions in partially observable monte-carlo planning: A rule-based approach

G Mazzi, A Castellini, A Farinelli - arxiv preprint arxiv:2012.12732, 2020 - arxiv.org
Partially Observable Monte-Carlo Planning (POMCP) is a powerful online algorithm able to
generate approximate policies for large Partially Observable Markov Decision Processes …

Learning state-variable relationships in POMCP: A framework for mobile robots

M Zuccotto, M Piccinelli, A Castellini… - Frontiers in Robotics …, 2022 - frontiersin.org
We address the problem of learning relationships on state variables in Partially Observable
Markov Decision Processes (POMDPs) to improve planning performance. Specifically, we …

[PDF][PDF] Policy Interpretation for Partially Observable Monte-Carlo Planning: a Rule-based Approach.

G Mazzi, A Castellini, A Farinelli - AIRO@ AI* IA, 2020 - ceur-ws.org
Abstract Partially Observable Monte-Carlo Planning (POMCP) is a powerful online algorithm
that can generate online policies for large Partially Observable Markov Decision Processes …

Active generation of logical rules for POMCP shielding

G Mazzi, A Castellini, A Farinelli - Proc. of the 21st International …, 2022 - iris.univr.it
We consider the popular Partially Observable Monte-Carlo Plan-ning (POMCP) algorithm
and propose a methodology, called Active XPOMCP, for generating compact logical rules …

Bayes-Optimal, Robust, and Distributionally Robust Policies for Uncertain MDPs

J Li, Z Dong, S Han - 2024 - indigo.uic.edu
We explore the performance of different policy-making strategies within the framework of
uncertain Markov decision processes (uMDPs). In an uMDP, the agent interacts with an …