DireenTech worked with the US Air Force Academy's Center for Unmanned Aircraft Systems Research (ACUASR) on using General Purpose Graphics Processing Units (GPGPU) in UAVs for fast autonomous target identification and GPS-denied navigation. A further objective was a general purpose UAV software platform, built around those GPGPU capabilities, that makes it easy for researchers and cadets to rapidly prototype software for UAV missions.
The research included the following.
GPGPU-enabled target identification
- Research, evaluate, optimize, and extend known image processing and computer vision algorithms for enhanced, autonomous target identification using GPGPUs.
- Perform autonomous target classification, such as identifying cars, people, animals and buildings within an image.
- Determine target velocity across multiple images relative to fixed features.
- Incorporate GPGPU image processing into the general purpose UAV software platform.
GPGPU GPS-denied navigation
- Establish GPS-denied navigation techniques using GPGPUs.
- Incorporate GPS-denied navigation into the general purpose UAV software platform.
General purpose UAV software framework
- Build a general purpose software framework for controlling UAVs in a wide range of ISR (intelligence, surveillance, and reconnaissance) missions, on fixed wing and quadcopter/hexacopter airframes.
- Research and develop a behavioral processing system that makes it easy to configure for a wide range of missions.