Project Proposal II

Crowd Flow Statistical Modeling in the Unreal Development Kit

Eric Baker and Brian May

Computer Graphics II; 4005-762

Reynold Bailey

http://bam1286.blogspot.com/

Summary: We plan to use the UDK to model crowd behavior in a given space. The system will use flocking and crowd AI algorithms to control actors inside of a pre-generated area. By tracking the path and interaction of the actors inside of the space, we can generate a "heat" map that will show where the highest densityof actor interaction occurs. To begin with, we will focus mainly on collisions as the actors attempt to navigate a number of obstacles.

Objectives: To generate a statistical model that will project actor behavior in a given space based on heatmap data from previously generated data.

System and Software: We plan to implement the system on the UDK in Windows 7.
A breakdown of project components: We will create an AI module for actors in Unreal that will implement specific flocking algorithms to allow them to navigate obstacles in a given space.

Project Responsibilities
Brian will implement heatmap generation based on actor interaction in the game world.
Eric will concentrate on implementing the flocking algorithms inside the engine.

Project Timeline

• Week 5 - setup environment
• Week 6 - Initial Algorithm implementation / Heatmap generation
• Week 7 - Refine and identify problems with algorithm and statistics gathering
• Week 8 - Finalize implementation and begin to gather data
• Week 9 - Present Findings

Final Presentation: We plan to demonstrate the system in real-time running on a machine in-class. Also we plan to have a short video / powerpoint that will present our findings and the implications of the system in terms of modeling spaces for future actor interactions.