Accelerating Route Planning and Collision Detection for Computer Generated Forces Using GPUs
Abstract
We present algorithms to accelerate route planning and collision detection for computer generated forces. Our algorithms exploit the parallel computing capability of Graphics Processing Units (GPUs) along with their ability to perform geometric culling. We combine the GPU accelerated computations with exact intersection tests on the CPU. Our approach supports dynamic terrains and multiple feature intersections in parallel. Our technique has been integrated into OneSAF block D build 24. Our route planning technique is a 30x - 50x speedup and has demonstrated an overall speedup of 10x. Our collision detection code is a 5x - 10x speedup over existing collision detection techniques.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 2006
- Accession Number
- ADA481888
Entities
People
- Angel Rodriguez
- Brian Salomon
- David Tuft
- Dinesh Manocha
- Maria Bauer
- Michael Macedonia
- Ming C. Lin
- Naga Govindaraju
- Russell Gayle
Organizations
- University of North Carolina at Chapel Hill