Rolling horizon co-evolution in two-player general video game playing

Artificial Intelligence for General Video Game Playing (GVGP) is challenging not only because agents must adapt to a range of different games, but they must also make decisions within the time constraints of real-time video games. The General Video Game Artificial Intelligence framework (GVGAI) is a popular framework for GVGP. It features a two-player track where two agents play a game together, either competitively or cooperatively, which poses the additional challenge of considering another player. Commonly, agents only consider their own moves in these two-player games. In this paper we discuss and assess Rolling Horizon Co-evolutionary Planning (a modification to Rolling Horizon Evolutionary Algorithms) for two player GVGAI. We present experimental results on its effectiveness against other agents playing GVGAI games and show that co-evolution can improve results compared to a RHEA agent.

Metadaten
Author:	Charles Ringer, Cristiana Pacheco, Georgiana Cristina Dobre ORCiD GND, Diego Perez-Liebana
Frontdoor URL	https://opus.bibliothek.uni-augsburg.de/opus4/112658
ISBN:	978-1-7138-2942-3OPAC
Parent Title (English):	AISB Convention 2021: Communication and Conversations, 7-9 April 2021, online
Publisher:	Society for the Study of Artificial Intelligence and Simulation of Behaviour
Place of publication:	Bath
Type:	Conference Proceeding
Language:	English
Date of Publication (online):	2024/04/23
Year of first Publication:	2021
Publishing Institution:	Universität Augsburg
Release Date:	2024/04/23
First Page:	18
Last Page:	23
Institutes:	Fakultät für Angewandte Informatik
	Fakultät für Angewandte Informatik / Institut für Informatik
	Fakultät für Angewandte Informatik / Institut für Informatik / Lehrstuhl für Menschzentrierte Künstliche Intelligenz
Dewey Decimal Classification:	0 Informatik, Informationswissenschaft, allgemeine Werke / 00 Informatik, Wissen, Systeme / 000 Informatik, Informationswissenschaft, allgemeine Werke

Open Access