Army Wants Quick Virtual Terrain


Michael Peck

May 22, 2012

Defense News

With so much hard labor needed to create virtual terrain for simulations, the U.S. Army is hoping that someone can come up with a quickie terrain generator for its land-attack missile simulations. Though simulations are often billed as a fast and low-cost method of training and mission planning, generating the underlying terrain models is a process built upon brute-force computing and copious amounts of human sweat, which eats up time and dollars.

The Small Businees Innovation Research project, titled "Rapid Scene Creation for Multispectral Terrain Signature Models and Simulaitons," calls for an automated technique or system that can quickly create earthen background terrain databases for integration into missile flight simulations.

Quick, high-fidelity terrain generation could also help with mission planning, unmanned aircraft approach routing, and civilian applications, such as natural resource monitoring and border control. Scene sizes range from 0.5-by0.5 kilometers to 10-by-10 kilometers, with terrain characterized by topography, features such as trees and roads, and EO/IR characteristics.

However, the army faces a tough challenge, said Nick Giannis, vice president of research and technology for Presagis, makers of the Terra Vista terrain software.

"The difficulty lies in the fact that they are trying to simulate infrared and optical sensors like FLIR [Forward looking infrared] and NVG [night vision goggles], and to this properly, you need to know the kind of material the terrain is made of, not just how it looks," Giannis said.

When using those types of sensors, the way the scene appears depends on the material, which reflects light or stores heat in different ways.

"For example, in order to properly simulate the appearance of a road crossing the desert as seen through the eyes of a helicopter pilot using FLIR, the simulated terrain must store the face that the road is made of tar and the desert is made of sand," Giannis said. "It is not enough to have it look right using the naked eye."

Automating this process will be a particular hurdle, he said.

"Having computers recognize roads automatically is hard enough. Adding the complexity of being able to tell what material that road is made of is even more difficult."

See completion version of Simulation Highlights, July 2012