The Simulator Revolution

Dec. 1, 1989

When the Apollo astronauts first landed on the moon some twenty years ago, they were thor­oughly prepared for that first step because they had rehearsed the mis­sion hundreds of times in simulators back on earth.

Mission rehearsal was the key to the success of the Apollo program. Now it is becoming critical to suc­cess—and survival—in the increas­ingly demanding world of tactical warfare. Fortunately for the US Air Force, supporting technologies are keeping pace with the challenge.

Apollo astronauts trained rigor­ously for two years in a mission sim­ulator that would be considered primitive by today’s standards. They “landed” a replica of their lu­nar module, using actual flight con­trols, on a simulated area of the Sea of Tranquillity known as a model board. They viewed this subscale world out the window via closed-circuit television. As the astronauts manipulated their controls, the TV camera moved correspondingly to give them a realistic sense of mo­tion.

Until the computer revolution hit the simulation business with gale force within the past decade, that’s all mission-rehearsal simulators were: TV cameras, model boards, and replicas of flight vehicles. Now the outside world is being re­produced digitally in the bowels of computers and displayed to the trainees in a way that allows them to interact with a broad range of stress-inducing situations.

This new technology is called computer-generated imagery (CGI), and it is the foundation for new training methods with sufficient re­alism to prepare today’s warriors for tomorrow’s challenges.

A Broad Geographic Sweep

Unlike TV model boards, CGI simulators can provide trainees with pictures of large geographic areas (including the routes to and from targets as well as the targets themselves) in which all the threats are accurately located with the aid of timely intelligence data. The as­tronauts could be reasonably confi­dent that there wouldn’t be anybody on the moon shooting at them, but that would not be the case for Spe­cial Operations Forces on missions to such areas as the Middle East.

Furthermore, in the increasingly threatening environment of elec­tronic warfare, mission success will depend on sensor data from outside the narrow visual portion of the spectrum. These data can also be computer-generated during mission rehearsals. So can fog, smoke, and haze. Just as the new sensor suites are intended to give fighter aircraft all-weather, day/night capabilities, their supporting mission-rehearsal simulators must do likewise.

George H. Branch III, manager of military marketing at General Elec­tric’s Simulation and Control Sys­tems Department in Daytona Beach, Fla., sees a trend toward greater re­liance on nonvisual data in both training and actual missions. Ten years ago, the out-the-window view amounted to 100 percent of tactical-warfare simulation, he says. Today it’s seventy-five percent and drop­ping. He sums up the situation suc­cinctly: “There’s more avionics to simulate.”

These nonvisual data, which oc­cupy much larger portions of the electromagnetic spectrum, include forward-looking infrared (FLIR) and narrow-field infrared, synthetic aperture radar, night-vision gog­gles, and low-light-level TV (LLLTV). This increased data flow requires sensor fusion techniques to funnel vital information to the pilot [see “Sensors Across the Spectrum,” November ’87 issue] in both the operational vehicles and the mission-rehearsal simulators.

Fifty Billion Instructions

That, in turn, increases the need for computer power to run today’s state-of-the-art CGI simulators. For example, the MH-53J helicopter weapon system trainer, which GE is developing for the Air Force’s Spe­cial Operations Forces, uses a com­bination of general- and special-purpose computers with processing speeds ranging from ten to fifty bil­lion instructions per second, ac­cording to Mr. Branch. That is much faster than even the most powerful supercomputers of today, although the two classes of machines aren’t quite comparable because of the specialized nature of simulation computing.

The data-storage requirements are equally demanding. To simulate a 300,000-square-mile area of the United States used for Air Force training exercises (essentially from Arkansas to Kentucky and parts of California), GE used four 300­-million-byte disk storage devices. To simulate the 3.6 million square miles of the fifty states would re­quire twelve times that amount. Of course, for mission rehearsals the areas to be simulated would be mostly in the Eastern Hemisphere, and the database for that is available from the Defense Mapping Agency and from what are known in the trade as “national technical means.”

The visual fidelity of CGI simula­tors is good and getting better, to the point where further improvements may not be necessary. As a rough measure of the capability of the human eye, if the normal field of view is digitized, it amounts to about a million pixels (picture ele­ments) of direct vision and roughly another million pixels of peripheral vision.

Today’s CGI simulators update the scene sixty times a second to give the illusion of reality. The human eye cannot sense individual pictures at rates greater than twenty-four a minute; therefore, that is the rate used in motion pic­tures (although each frame is pro­jected twice to eliminate the jerky motion of the early silent films).

This rate, providing a further smoothness of motion, is essential in interactive mission simulations because conflicting visual cues can cause motion sickness among the trainees.

Thus the computational require­ment for CGI is dictated by the need both to provide at least a million digitized picture elements per scene and to do it sixty times a second. That’s where today’s computers built out of very-large-scale integra­tion (VLSI) components have taken over, muscling out TV model boards in the process. “The picture quality is there,” says Mr. Branch. “No more pixels are needed.”

Antithesis of “Simnet”

This approach of high fidelity, relatively high costs, and limited inter­action for simulators based on powerful stand-alone central com­puters can be thought of as the an­tithesis of the Defense Advanced Research Projects Agency’s Simnet (simulator network) approach. Simnet uses low-cost distributed com­puters to produce maximum inter­action among participants in train­ing exercises, but at this point it is capable of only relatively crude graphics [see “Planet Simnet,” Au­gust ’89 issue, p. 60]. It is reason­able to expect that, in the future, these approaches could converge to create even more powerful simula­tors.

According to Michael R. Willmore, a staff scientist at Link Flight Simulation, Binghamton, N. Y., a division of Toronto-based CAE Industries, effective mission rehearsal depends on countering three kinds of uncertainty: situa­tional uncertainty, probabilistic un­certainty, and operational uncer­tainty.

Situational uncertainty applies to the purely physical nature of a re­gion where the conflict is to be mod­eled, essentially terrain and weath­er. Probabilistic uncertainty in­cludes the capabilities of the weap­ons that all the participants bring to the battlefield: system perfor­mance, reliability, probabilities of hit and kill, even electronic signa­tures. Both of these are well within the realm of current simulation technology, Dr. Willmore maintains.

The outlook is not so bright for operational uncertainty. Dr. Willmore calls it the most difficult as­pect of warfare to simulate or even account for in reality. It is the result of how cohesively the command structure is organized, how efficient the control processes are in direct­ing force responses on the battle­field, and the connectivity strength of communications systems in pass­ing essential information among the entire command control and com­munications (C3) architecture.

“It is pointless to design a static threat simulation for mission re­hearsal that can only record and play back one presupposed set of conclusions about the mission en­vironment or what the conflict should look like during mission re­hearsal,” Dr. Willmore states. “Such ‘tactical’ simulations, cre­ated by writing scripts from a set choreography, cannot possibly re­spond to the dynamics generated by a single participant, let alone sev­eral others who may be operating together as a mission unit.

“Instead, mission rehearsal should serve as an adjunct to the final mis­sion planning activity that occurs just prior to executing tactical mis­sions in reality,” he continues. “Participants explore the planned missions by asking themselves, ‘What if we did this?’ and ‘What if the enemy does that?’ and ‘What if this happens?’ and the entire litany of other questions designed to bet­ter prepare themselves for the un­certainty at hand.”

High Costs—For Now

Then there’s the issue of costs. Simulators aren’t cheap. GE’s MH-53J system, for example, is projected to cost more than $30 mil­lion. But they are getting cheaper, at least on a cost-per-function basis. Through the use of VLSI compo­nents (and soon, it is hoped, trans­portable software), simulators are getting smaller, cheaper, and easier to support. Mr. Branch estimates this price decline at about ten per­cent a year, but he cautions that sim­ulator prices are likely to remain steady because the military custom­ers are likely to opt for increased performance instead of lowered sys­tem costs.

A rule of thumb in the industry is that the customer will pay about ninety percent of the unit cost of the aircraft for its simulator. In the case of the Air Force’s Advanced Tactical Fighter (ATF), which has a pro­jected $35 million program unit cost, that means a likely ceiling price of close to $32 million for the simulator.

Development of the simulators for ATF, as well as those for the X-30 National Aerospace Plane, the aircrew training system for the Spe­cial Operations Forces, and the up­grade of the F-16 simulators, are all managed now out of the System Program Office for Training Devices (still referred to as SIM/SPO) under Col. Wayne Lobbestael at Aero­nautical Systems Division, Wright-Patterson AFB, Ohio.

This is a departure from past Air Force practice, in which the simula­tor efforts had been under the SPO managing the weapon system devel­opment. The Army and Navy have centralized their simulator develop­ment and procurement under the Program Manager for Training De­vices (PM-TRADE) and the Naval Training Systems Center, respec­tively, both located in Orlando, Fla. Centralizing the simulator effort re­moves it by at least one step from the budgetary pressures that nor­mally afflict weapons development programs—a distinct advantage.

Navy, USAF Take Different Paths

Because of the differing natures of their tactical air missions, the Air Force and Navy have taken differ­ent approaches to flight simulation. Since Navy fighters customarily op­erate off the decks of aircraft car­riers, the Navy early on recognized the benefits of simulation to reduce the number of risky carrier opera­tions. A classic example is an engine flameout during a carrier landing, something no pilot wants to practice in a real aircraft.

The Air Force has not felt such a need for flight simulators and did not introduce visual simulation until the recent F- 16 upgrade program re­cently won by Evans & Sutherland of Salt Lake City, Utah. Dave Ec­cles, manager of strategic planning at E&S, describes the new F-16 sim­ulators as relatively small field-of­-view devices capable of simulating takeoffs and landings and some mis­sions—but not traditional full-mission simulators. These are also relatively low-cost, estimated at about $1.5 million apiece.

But Mr. Eccles sees other forces at work that may win further cus­tomer acceptance of flight simula­tors. His company recently re­ceived a contract to supply at least six low-level flight trainers for the West German Tornado fighter, and this may be a bellwether for future procurements. Just as one of the purposes of DARPA’s Simnet is to prevent tanks from tearing up farm­land and causing intolerable traffic jams in West Germany, simulators for tactical aircraft in the NATO en­vironment can be a force for better relations among NATO allies.

Looking beyond these current ap­plications of flight simulators, Mr. Branch of GE traces the impact of size reduction made possible by new electronic components. GE’s original Compu-Scene II system, introduced in 1980, consisted of twenty-six cabinets, each standing about six feet high and weighing 900 pounds. Compu-Scene V, intro­duced at this year’s Paris Air Show, dropped that to six cabinets, and Mr. Branch says the next goal is to get an entire simulator into a single cabinet.

At 900 pounds per cabinet, the simulator could easily be installed on board an aircraft the size of a USAF C-5 transport to permit em­bedded training during normal flight operations. Another order of magni­tude reduction, down to ninety pounds, would put that capability within reach of the ATF.

The Totally Enclosed Aircraft

Given the increasing importance of nonvisual sensor data, future de­rivatives of today’s flight simulators might entirely replace the out-the­-window view. Submarine com­manders have been doing this for years. They rarely peer through periscope eyepieces anymore; the sensor data are funneled to them through a variety of mast-mounted devices and displayed in the sub­marine control center on television screens. This enables submarines to reduce their visibility to enemy forces.

In the case of high-performance fighters, it might be more efficient for the pilot to be in a supine posi­tion monitoring the sensor data over CCTV during periods of high G-forces. This approach could elimi­nate the traditional cockpit entirely, which would be valuable in reduc­ing the aircraft’s radar cross sec­tion. Pilots are already overly task-loaded with through-the-window data, and the use of sensor fusion could eliminate extraneous infor­mation. The value of sealing off the aircraft in a nuclear environment is obvious.

Taken together, these potential capabilities of CGI give this tech­nology the edge for a variety of fu­ture applications. TV model boards put Americans on the moon and per­formed many other valuable func­tions, but today their importance has shrunk to what Mr. Eccles of E&S calls the equivalent of HO-scale railroad models.

John Rhea is a free-lance writer living in Woodstock, Va., who specializes in military technology issues. His most recent article for AIR FORCE Magazine, “Silicon’s Speedier Cousins,” appeared in the November ’89 issue.