Commercial gear sets the standard for military push into embedded training

Aug. 1, 2004
ORLANDO, Fla. — The military's hopes and plans for embedded training rely heavily on continued growth and competition in the commercial market.

By J.R. Wilson

ORLANDO, Fla. — The military's hopes and plans for embedded training rely heavily on continued growth and competition in the commercial market. In fact, military officials who are in charge of procuring embedded-training subsystems are using well-known commercial-grade systems as informal standards.

Any industry bid to provide embedded-training systems that are not at least as good, technologically, as the latest version of the PlayStation or Xbox commercially available video games will have no chance with the military, says Chris Marzilli, vice president at General Dynamics C4 Systems in Taunton, Mass.

Embedded training involves devices on weapons platforms such as aircraft, warships, or main battle tanks that artificially stimulate onboard sensors such that weapons behave as if they were operating in a real battle. The idea is to enable systems operators to train on the system they would actually use in battle.

Military embedded-training systems also must be able to upgrade to faster speeds and higher- performance graphics without falling victim to obsolete chips or boards. That essentially means each new U.S. Army Stryker brigade or Future Combat System (FCS) buy must incorporate new technology that then will retrofit into already fielded platforms.

"There also is a desire to do things over direct pipes, so you can even remove and replace the motherboard to get a direct pipe from the CPU to the graphics module," Marzilli says. "The box itself is custom for the Stryker vehicle — types of connectors, cable routing, paths of external interfaces to the displays, etc. — but the internals have been designed as very modular and upgradeable. So as FCS comes on, that should still be fresh and relevant. Everything is based on industry standards, providing much higher confidence that it will survive lots of evolution."

Oasis Advanced Engineering in Auburn Hills, Mich., is among those focusing on developing new training capabilities for relatively old systems. One such effort is the M1A2 Abrams SEP (System Enhancement Package) Tank Embedded Training Technology.

"We have designed a single LRU (line replaceable unit) — roughly 8.5 by 9 by 14 inches, weighing about 40 pounds," Oasis President Carl Hobson says. "Everything is conduction cooled, full MILSPEC."

Instead of using vehicle-modeling software and other packages that were not designed to be embedded or work with other models, Hobson and his team took appropriate material from the library, repackaged it, and dropped it into the vehicle with all interfaces matched with the actual vehicle signals. Instead of getting information from vehicle-modeling data, the system gets information from the actual tank. That enables embedded training as well as mission rehearsal.

"We are a systems integrator. We don't actually build the board housing, and everything in that box can be bought off the shelf," Hobson says. "We had to take Windows XP Embedded and Linux (operating systems), and optimize them to work in this environment, then rewrite all the device drivers."

Hobson integrated embedded software along with the government-owned advanced-gunnery-training-system (AGTS) software into the box. "We also had to modify actual tank software, which we were able to do based on 15 years of working with the tank. We enhanced that software so in addition to normal combat and maintenance modes, it can be put into training mode, exercising the actual vehicle software. In the future, tank software and all embedded-training software can be released as a package, needing only to include the LRU to make it functional," Hobson says.

Holistic approach

At Lockheed Martin Electronic Systems Advanced Technology Laboratories in Cherry Hill, N.J., work is in progress on several advanced embedded-training prototypes, most for the Office of Naval Research (ONR) in Washington. Those efforts focus on human-performance improvement technologies, and automated and semi-automated performance measurements that can provide feedback during and after an exercise — what lead engineer Dr. Peter Bilazarian calls intelligent embedded training, intelligent diagnosis, and intelligent feedback.

"This holistic approach gives users an ability to understand what they are doing as they are going along and to review afterward," he explains. "Rapid deployment and multiple missions require a much more intelligent and adaptive system. We complement simulation, advanced simulation, and ET (embedded training) by creating capabilities allowing them to automate and semi-automate (observer-in-the-loop) feedback on objectives and performance."

John Anderson, manager of the distributed processing lab at Lockheed Martin Electronic Systems, says the Navy's efforts to improve efficiency and effectiveness — including the number of people necessary to support deployed embedded training — are driving initiatives.

"When instructors were watching people train in the past, they could see if they hit the target. But in command-and- control (C2) exercises, you have people talking into a mike, making entries with keyboards and trackballs, pushing buttons, and so on," Anderson says.

"We tried to more rigorously capture those keystrokes and button actions and relate those to meaningful actions, then compare what the operator did with what an expert would do in response to a critical-scenario event. The computer does a better job of capturing the data, and we need fewer human instructors. We found one instructor could handle and evaluate up to four operators in a C2 environment."

The next step is to apply all that to the training and evaluation of several distributed teams, not only within a given service but also in joint and coalition scenarios. Different objectives and approaches to training complicate this, however.

Maturing technologies

"There are a lot of maturing technologies coming out that are slowly being applied to this process," Anderson says. "If you have two different evaluation paradigms, one based on tasks and the other on objectives, there is a search under way to develop ontologies to promote interoperability and communication. Our recent activities have raised the obvious need for interoperability, and a lot of energy is being applied by all the services to become more interoperable internally as well as with other services."

An important aspect of that is to go beyond measuring outcomes or even whether someone made a correct decision, adds Lockheed Martin's Bilazarian. The idea is to look at the process from an individual, team, and multiple team perspective to see if decisions — correct or not — were made using the correct processes. Also important is getting away from designing a system, then forcing people into it; instead, the new emphasis is on accommodating humans as integral parts of the system, so it is not necessary first to train someone to use the trainer.

Embedded training also is a candidate for the fast-growing and increasingly important area of unmanned vehicles in the air, on the ground, and underwater. In combat, these systems are likely to operate over unfamiliar terrain, so a key capability is the ability to create and update terrain databases quickly.

"Right now, terrain databases are the determining step in simulation. They take a lot of money and time to build," says Garth Smith, president of MetaVR in Brookline, Mass. "Our tools can build databases pretty quickly. As new data comes in, you can update an existing database or rapidly create a new one. You don't have to update the entire database at one time; you can deal with manageable pieces, one at a time, to handle just those elements that actually have changed. The technology exists to do this; they just need to get it out into the field."

The future

"Eventually, you could have real-time visual imagery distributed to every commander or even individual soldier in the field," says Smith. "It could even be used by commanders to send specific tasks to those units in the field, with the commander pointing to where he wants them to go in the 3D visual environment being generated in real time using incoming data from satellites, UAVs, and other intelligence resources."

Smith says one application may tie to training for military operations in urban terrain (MOUT), which already is expanding from a few large fixed-base specialty facilities, such as Fort Polk, La., to home base and mobile MOUT-training facilities. Instead of spending millions of dollars to build a 3-D MOUT database, the same system building databases for operational field applications can build one for MOUT use. It also could replay a MOUT exercise in 3-D, showing just what happened, moment by moment, or provide refresher exercises online when an actual MOUT site is not available.

In addition to future systems such as FCS and even Stryker, United Defense LP in Arlington, Va., is on its third generation of providing a form of embedded training for the legacy Bradley, which Mark Russell, manager of the Simulation, Training & Systems Group, says is primarily driven by commercial video computers and cards.

"The technology could be fully embedded on a MILSPEC VME card in six to eight months," he predicts. "We're currently under contract from PM-Combat Systems to build one prototype demonstrator for delivery in February, which I believe they plan to send to Iraq for field evaluation."

The biggest challenges in putting a trainer into a combat vehicle are cost-effectiveness and durability, especially if the trainer is being retrofitted into an existing platform.

"A lot of the capability in today's trainers is driven by commercial computers and components, so you have to do some engineering analysis to figure out how to put that in a combat vehicle without it breaking down on the battlefield and do that cost-effectively," Russell says. "One game plan is to share the memory and processing power with other applications not used in the vehicle when training is in effect.

"On the Bradley, we are using five cards that can be installed in the turret processor or the hull-processor unit, which are basically LRU black boxes with open slots. Some other legacy vehicles also have open card slots, so it could be relatively simple to do the same. These are basically standard computer cards, but MILSPECed because of the conductive cooling — where a standard CPU card is air cooled by fans — and for shock and vibration. The only difference would be in the video chips; you would need high-end commercial video chips to run the graphics, but those same chips typically would be needed to run streaming video from a tactical UAV, so this could be a dual-purpose card."

However it is accomplished — integrated into the operational system, as an add-on module or by linking to an external device — taking training to the soldier in the field is a concept whose time has come as an enabler of true combined-arms training, according to Kelley.

"If I'm a company commander and get new personnel or new technology, I have the ability, in the assembly area, to train my company on a truly ET system and run them through likely scenarios we will be involved in before we actually engage in battle," he says. "That is very powerful." ..

Voice your opinion!

To join the conversation, and become an exclusive member of Military Aerospace, create an account today!