The MADP Toolbox: An Open-Source Library for Planning and Learning in (Multi-)Agent Systems



Oliehoek, Frans ORCID: 0000-0003-4372-5055, Spaan, Matthijs TJ, Robbel, Philipp and Messias, Joao V
(2017) The MADP Toolbox: An Open-Source Library for Planning and Learning in (Multi-)Agent Systems. In: The AAAI Fall Symposium on Sequential Decision Making in Intelligent Agents, Arlington, VA, USA.

[img] Text
Oliehoek15SDMIA.pdf - Unspecified

Download (127kB)

Abstract

This article describes the MultiAgent Decision Process (MADP) toolbox, a software library to support planning and learning for intelligent agents and multiagent systems in un- certain environments. Some of its key features are that it sup- ports partially observable environments and stochastic tran- sition models; has unified support for single- and multiagent systems; provides a large number of models for decision- theoretic decision making, including one-shot decision mak- ing (e.g., Bayesian games) and sequential decision mak- ing under various assumptions of observability and coopera- tion, such as Dec-POMDPs and POSGs; provides tools and parsers to quickly prototype new problems; provides an ex- tensive range of planning and learning algorithms for single- and multiagent systems; and is written in C++ and designed to be extensible via the object-oriented paradigm.

Item Type: Conference or Workshop Item (Paper)
Uncontrolled Keywords: software, decision-theoretic planning, reinforcement learning, multiagent systems
Subjects: ?? QA76 ??
Depositing User: Symplectic Admin
Date Deposited: 26 Nov 2015 09:12
Last Modified: 16 Dec 2022 02:35
Related URLs:
URI: https://livrepository.liverpool.ac.uk/id/eprint/2039201