|
Guest >> Sign In |
|
|
DIRECTED
ENERGY
PROFESSIONAL
SOCIETY
Abstract: 25-Symp-013
|
|
|
|
UNCLASSIFIED, PUBLIC RELEASE
Graphics Processing for Real-Time Directed Energy Tracking Systems
We propose a graphics processing unit (GPU)-based computing and processing for improving the usability and maintainability of a Directed Energy (DE) tracking system, while reducing development time and costs with respect to current field programmable gate array (FPGA) approaches. The computational demand of image processing and tracking in advanced directed energy applications exceeds the capabilities of central processing unit (CPU)-based methods, leading to the adoption of the FPGA approach. The FPGA approach, while fast enough for DE tracking, requires specialized programmers, long development times, and high-cost, proprietary tool chains and libraries. Our approach attempts to mitigate these downsides by investigating integration of GPU-based acceleration in targeting and tracking. This approach enables generalist programmers to develop the system using C++ and the C-like languages through NVIDIA’s CUDA platform, such that skilled programmers from a standard computer science program can contribute. The approach also enables smooth integration with the Linux operating system, allowing usage with industry standard software for activities including software engineering, project management, debugging, and hardware integration. We integrated our GPU approach using an NVIDIA Jetson Orin Nano board for real-time processing of attached USB3 machine vision cameras and conducted preliminary measurement of performance. This approach yields exceptional performance, demonstrating the capability to execute 2500 cross-correlation operations per second on 256 by 256 frames, output by the camera at 2500 frames per second. This capability is beneficial in both real-time object tracking and laser systems where rapid target acquisition is critical. The performance achieved through GPU acceleration validates its potential as a solution to the need for high-speed real-time image processing applications, thus supporting capabilities in advanced directed energy platforms.
The views expressed are those of the author(s) and do not reflect the official views of the United States Air Force, Department of Defense, United States Government, or United States Space Force. Mention of trade names, commercial products, or organizations do not imply endorsement by the U.S. Government.
Approved for public release; distribution is unlimited. Public Affairs release approval # AFRL-2025-1064
UNCLASSIFIED, PUBLIC RELEASE
|