Beyond playing to win: Diversifying heuristics for GVGAI

Abstract

General Video Game Playing (GVGP) algorithms are usually focused on winning and maximizing score but combining different objectives could turn out to be a solution that has not been deeply investigated yet. This paper presents the results obtained when five GVGP agents play a set of games using heuristics with different objectives: maximizing winning, maximizing exploration, maximizing the discovery of the different elements presented in the game (and interactions with them) and maximizing the acquisition of knowledge in order to accurately estimate the outcome of each possible interaction. The results show that the performance of the agents changes depending on the heuristic used. So making use of several agents with different goals (and their pertinent heuristics) could be a feasible approach to follow in GVGP, allowing different behaviors in response to the diverse situations presented in the games.

General Video Game AI: A Multitrack Framework for Evaluating Agents, Games, and Content Generation Algorithms

“…Guerrero et al [81] explored five GVGAI agents using four different heuristics separately on playing twenty GVGAI games, allowing different behaviors according to the diverse scenarios presented in the games. In particular, this work explored heuristics that were not focused on winning the game, but to explore the level or interact with the different sprites of the games.…”

Mek: Mechanics Prototyping Tool for 2D Tile-Based Turn-Based Deterministic Games

“…The General Video Game Artificial Intelligence (GVGAI) framework [1] is a system created with the goal of encouraging AI researchers to focus their attention on general problem solving methods in order to allow them to be more readily transferable to new domains. Through defining a common interface to a large number of distinct games, the framework facilitates exploration of a number of game development areas, including: general AI agents playing unseen games [7], generating levels [8] and, most relevantly, generating game rules [9].…”

Ensemble Decision Systems for General Video Game Playing

“…3) Knowledge Discovery Heuristic (KDH): Its goal is interacting with the game as much as possible to trigger interactions, and new sprite spawns. For these experiments, this heuristic has been updated regarding the one in [16], so winning the game is always rewarded instead of penalized. 4) Knowledge Estimation Heuristic (KEH): Its goal is interacting with the game to predict the outcomes of the interactions between the different elements of the game, related to both the victory status and score modifications.…”