TECHNOLOGY AREA(S): Human Systems
OBJECTIVE: Develop the capability to rapidly generate lightmap based models to enhance infrared capabilities in game engines/image generators to support C4ISR personnel training.
DESCRIPTION: State-of-the-art Command and Control, Computers and Communications, Intelligence, Surveillance and Reconnaissance (C4ISR) training research requires rapid generation of synthetic virtual training vignettes to enable rapid response to requirements of the future fight. Existing efforts at the National Aeronautics and Space Administration (NASA), Air Force Research Laboratory (AFRL), and U.S. Army Research, Development and Engineering Command (RDECOM) enable generation of synthetic terrain using real-world imagery. However, these efforts only produce terrain. A critical shortcoming is the inability to render realistic infrared representations in real-time. C4ISR subject matter experts have stated that rendering physics-based sensor models, especially but not exclusively infrared, is an essential capability. A key capability required for physics-based sensor models is to develop cumulative temporal energy maps that models accumulated energy stored by a surface. This requires dynamic computation because moving entities shadow areas, allowing energy to dissipate. One potential approach to the development and implementation of physics-based sensors models is the modification of lightmap rendering technology to create temporal energy maps. State-of-the-art game engines have advanced, photo-realistic lightmap technology that affects the appearance of modeled environments. One key feature of lightmaps is the ability to customize the special effects of the light: its direction, intensity, how it reflects off what materials. These features should make it possible to render realistic infrared. However, while building terrain is a fairly fast effort even for very large areas, building lightmaps is exceptionally computationally intensive, requiring many hours for smaller tasks up to days or even weeks for more complex tasks. Conveniently, some game engines/image generators have an off-the-shelf distribution system to allow distributed builds on a High Performance Computing (HPC) system. Other novel approaches to the development of physics-based sensor models may also be considered. The scope of this effort is targeted at the development of a software to enhance real-time infrared capabilities within synthetic environments to support C4ISR training applications.
PHASE I: Research different physics-based sensor modeling methods and physics interactions, such as with materials. Develop methods of rendering infrared in image generator/game engine tools. Generate simple vignettes that demonstrate key rendering capabilities. Deliverable: A repository of images, movies, and interactive samples, demonstrating different approaches to rapidly generating realistic infrared imagery.
PHASE II: Configure and integrate the models into an existing image generation capability to support training effectiveness evaluation. Assess the capabilities of the prototype models in terms of fidelity and timeliness to meet the needs of C4ISR training. Deliverable: Configured and documented system in the C4ISR testbeds. Design and specify a stand-alone HPC infrastructure to enable local real-time energy map processing & streaming to support training RDT&E. Design document and bill of materials.
PHASE III: The immediate use case is directly applicable to the development of simulation-based training environments for the AF C4ISR domain. On the civilian side, this technology could advance the capability and fidelity of commercial gaming technologies.
1. Arnaud, R. & Jones, M.T. (2000) Image generator visual features using personal computer graphics hardware. IMAGE 2000 Conference.
2. Segl, K., Richter, R., Kuster, T. & Kaufmann, H. (2012) End-to-end sensor simulation for spectral band selection and optimization to the Sentienel-2 mission. Applied Optics, 51(4), 439-440.
KEYWORDS: Virtual Environment, Sensor Modeling, Training, Lightmap, Image