Lead: G. Michael Youngblood, Ph.D.
Team: Hunter Hale (Ph.D. Student), Priyesh Dixit (MS Student)

Mission:

To identify, develop, collect, and integrate a suite of tools and techniques for improving the understanding of interaction and behavior in all FPS/3PS games, incorporating knowledge as game artifacts, and learning how to advance game artificial intelligence through these mechanisms

Downloads:
PlayerViz v0.5

The PlayerViz tool is designed to allow interactive visualization of player trace data from one or more players. A player trace is represented as a series of spheres with connecting lines. The color of the spheres cycles through a rainbow colormap, blue to green to red, over time and the position of the spheres represents the position of the player. The player's orientation at that time is portrayed as an oriented line segment coming out of the sphere. A wireframe white sphere represents that the player found the goal and is usually at the end of a trace. A red wireframe sphere around the player means they lost health and a red solid sphere at the end of the line segment means they fired a shot.

HIIVVE v0.5

The Highly Interactive Information Value Visualization and Evaluation tool (HIIVVE) provides an interactive interface for finding information value of surfaces and was built using C++. The graphics rendering is done using OpenGL and the user interface is created in FLTK (Fast Light ToolKit). The view distance and the half-angle of the view frustums can be adjusted with a slider bar. The sorted surfaces list-box provides a list of all of the surfaces in order of information value once the Calculate Intersections button has been pressed. Rendering the lighting, XYZ axes, and the view frustums can be toggled with buttons. Due to the libraries used by the tool it is cross-platform and can run on Mac OS X, Linux, and Windows. However, this current version has only been tested on Mac OS X and Linux but we will make a Windows executable available very soon.

QCR AutoGen (Polygon) alpha

This application allows for the decomposition of areas containing both empty space and user defined obstructions(positive space) into a series of Qualitative Convex Regions(QCR). This is the first step to generating connectivity information between those regions to use for agent navigation. Normally this process would have to be done by hand and would take considerable time as well as generating a more triangular result. Our application will automatically generate a quad based decomposition which allows for easier navigation. Currently our application only handles Rectangular Positive Space regions, however work is progressing to allow for arbitrary shaped obstructions and even 3D generation.

SARGE (coming soon!)

Back to the Game Intelligence Group