Learning Avatar's Locomotion Patterns Through Spatial Analysis in FPS Video Games

Learning Avatar's Locomotion Patterns Through Spatial Analysis in FPS Video Games

Luis Alberto Casillas Santillan (University of Guadalajara, Guadalajara, Mexico) and Johor Ismael Jara Gonzalez (University of Guadalajara, Guadalajara, Mexico)
DOI: 10.4018/IJOCI.2018010103
OnDemand PDF Download:
No Current Special Offers


This article describes how current video games offer an extreme use of media fusion. Such construction implies a novel form of complexity regarding game control and active response from game to player. All of these elements produce deeper immersion effect in players. In order to perform a detailed supervision over this kind of game, additional controls should be included in game. Some of these controls are the moving and decision schemes. Authors believe that players move around virtual scenarios following some sort of pattern. Every player would have a specific pattern, according to his/her experience and capability to manage the gamepad layout. Current proposal consists in a 3D geometrical model surrounding player's avatar. Data unwittingly provided by the player, have elements to discover and, eventually, learn some gamers' patterns. The availability of these patterns would allow an improved game response and even the possibility of machine learning, as well as other artificial intelligence strategies. Every 3D game may include the model proposed in this paper, due to its noninvasive operation.
Article Preview

1. Introduction

Nowadays, video games offer extreme use of media fusion involving: video, audio, kinetics, sensing, etcetera. Such construction implies a novel form of complexity regarding game control and active response from game to the player. The trends in the gaming industry are aimed at providing realistic experiences. The first-person shooter (FPS) games are not excluded from this movement. Although there is a diversity of virtual environments, the present study is based on FPS. Complex scenarios in FPS imply challenges when trying to analyze and understand: player's behavior and game experience. Video games are sold as independent entities which cannot be accessed or manipulated due to copyright restrictions and patents' boundaries. Hence, a question emerges: how could it be performed an analysis of players when they are using a real video game?

Besides, there are not available standards helping to understand and unify the artificial intelligence (AI) approach used in different video games. In fact, every publisher would develop a different strategy regarding the AI matters. Thus, another question emerges: what is the regular behavior for autonomous entities in video games and how could it be analyzed?

One of the most difficult stages when complex video games are developed lies in the movement algorithm for non-player characters (NPCs), which are those animated entities moving around the game scenarios. Some of the NPCs are not dangerous meanwhile, the most frequent situation is that NPCs would try to attack or damage the player's avatar or the player's properties. Complex video games have a myriad of NPCs: sand, plants, water, clouds, insects, vehicles, robots, enemies, machinery, etcetera. These NPCs are active entities in video games with specific roles. NPCs operation would influence the whole game-playing experience. The movement algorithms for NPCs must consider interaction with other NPCs, as well as the interaction with fixed elements in scenarios and player's avatar.

Movement algorithms for NPCs are some of those restricted elements of video games, due to patents, industrial secrets, or copyrighting constraints. Just as secret recipes, these behavioral algorithms are highly involved in video games success. Unfortunately, there are no reference patterns that could be used for matching with actual behavioral scenarios. When NPCs movement is constantly based on the same algorithm, it makes a “monotone experience” for players. Such monotony leads to a no enjoyable experience from the game, implying a boring predictive behavior.

There are several threats when creating a dynamic experience every time the game is played. Such threats imply the need to expand gamers' experience. To do so, certain efforts are focused on improving the AI algorithms for NPCs. The goal is to add natural or even realistic movements to NPCs, including the fact that NPCs should be as restricted as players, e.g., NPCs do not see through the walls, NPCs need to reload their guns, NPCs have limited life/energy, NPCs must navigate space without leaping, etcetera. These aspects would provide the NPCs a new series of patterns that eventually are easily learned by players, but it does not necessarily represent a deeper experience on the game.

The present study is focused on taking advantage of players' movement, by analyzing their performance over a standard gamepad's layout. The goal is to discover how video gamers move around virtual scenarios and shed light on the search of locomotion patterns discovery. Every gamer would have a specific pattern, according to his/her experience and the capability to manage gamepad's layout, as well as his/her skills to play. Current proposal consists of a logging mechanism, which automatically collects raw data from gamers' activity. The gamer avatar is settled inside an imaginary polyhedral structure. This model allows collecting all the possible movement events, as well as the viewing events made by the player during the game experience. Hence throughout a synchronicity effort, involving game experience and the activity logs, different machine learning algorithms allow knowledge discovery involving gamers' behaviors. This experiment is aimed at developing a spatial structure enabled to model gamers' locomotion profiles, which will allow discovering behavioral patterns about user experience in video games, as well as the prediction of gamers' response for NPCs anticipation.

Complete Article List

Search this Journal:
Open Access Articles
Volume 12: 4 Issues (2022): 2 Released, 2 Forthcoming
Volume 11: 4 Issues (2021)
Volume 10: 4 Issues (2020)
Volume 9: 4 Issues (2019)
Volume 8: 4 Issues (2018)
Volume 7: 4 Issues (2017)
Volume 6: 4 Issues (2016)
Volume 5: 4 Issues (2015)
Volume 4: 4 Issues (2014)
Volume 3: 4 Issues (2012)
Volume 2: 4 Issues (2011)
Volume 1: 4 Issues (2010)
View Complete Journal Contents Listing