TRAINING AND SIMULATION
Real-World Imagery Sought For Military Training
When units have only days — and sometimes just a few hours — to prepare for a mission, not having the most up-to-date training databases can mean the difference between life and death.
“It’s a particular concern for that part of our customer community to have that concurrency because we’re asking special ops teams to go into an area and carry out a mission,” said Chester Kennedy, vice president of engineering at Lockheed Martin Global Training and Logistics. “We sure would like for them to be rehearsing that mission in a database that is very, very, very close to what they’re going to experience,” he said — not what it was six months or a year ago in that part of the world.
Training simulation databases are typically developed using satellite imagery. Engineers and graphic artists transform the two-dimensional photos into a 3-D representation of the terrain. It is a labor-intensive process that ensures all the elements in the environment — ranging from landscape features such as buildings, roads, rivers and trees to animated objects such as aircraft and troop transports — interact with the proper physical properties.
A typical database that covers a 5-kilometer-by-5-kilometer area of the world requires several “man-months” of effort to accomplish the conversion from a satellite photograph to a training database.
Automating the effort by using advanced image processing algorithms could accomplish the same goal in a matter of hours, Kennedy said. Those processors could take a downlink video from surveillance aircraft and turn it into a 3-D database just as easily as they might convert static 2-D satellite imagery.
Through this “run from source” process, engineers could take imagery that was collected in the morning by a jet fighter or by a UAV and have special operators training in the afternoon in a database developed around that, said Kennedy.
Lockheed has partnered with two firms to advance the concept. A Rochester, N.Y.-based company called Pictometry uses an airplane-mounted image capturing system that takes photos at multiple angles while flying at 3,000 feet to 5,000 feet.
“It’s a rapid way to collect data and process it with a high degree of density, texture and detail precisely because of those multiple camera shots,” said Thomas Burns, vice president of federal sales at Pictometry.
“Using satellite imagery typically requires a lot of rework with color balancing, correction, fusion, in order to make it usable in a database. Because of the fidelity of the images we capture, we’re able to bypass those issues and move quickly into rendering them for use,” he added.
The projection systems that display the images in a simulator also provide a challenge for database producers. Because image generators run different formats, they are not necessarily compatible with available databases. If one simulation contains a set of image generators but another one employs a different group of projectors, then two different databases are often needed.
“It’s a significant amount of work to convert a database from one image generator or projector to another,” said Kennedy.
With the “run from source” effort, the algorithms take care of the conversion and produce the output necessary for each projection system.
“A lot of the technology has been accelerated by the gaming industry and [advancements in] processing and graphics cards,” Kennedy said.