Embedded Training: Real-Time Computer Image Generation
Gregory F. Gustin
President, Paragon Graphics, Inc.
5460 Hoffner Avenue, Orlando, Florida 32812
Presented at Society of Photo-Optical Instrumentation Engineers (SPIE)
1987 Technical Symposium Southeast, 17-22 May 1987, Orlando, Florida.
Simulation devices have developed into critical tools for training operators of complex vehicles such as airplanes. As the costs of these vehicles have increased to exorbitant levels, so too have the costs of the simulation equipment increased. An alternative to producing and using more simulators is the introduction of embedded equipment to support training.
During the past two decades simulation devices have evolved to such a high degree of fidelity that, in many areas, there is literally no difference between the actual equipment and the simulator. When one ponders the sophistication of the latest generation of glass cockpits as found in modern aircraft, it might seem more difficult to provide high fidelity simulation. However, it is the presence of the on-board flight control computers (FCC), the high-speed multiplex bus (e.g. 1553) and associated "computer" peripherals, that actually provides the opportunity for such high fidelity simulation. Modern aircraft are frequently described as computer labs with "wings and motors." Therefore, if the on-vehicle computer network provides an environment that readily supports high fidelity simulation, does it not then stand to reason that this same system can provide on-board simulation to support training?
Aviation's Use of Embedded Training
The concept of embedded training is not, in itself, new. One of the most aggressive efforts to date has been the TRIAD program developed by Rediffusion. The TRIAD system was quite an unusual approach. Rather than incorporate ancillary simulation equipment within an actual vehicle, the TRIAD utilized an operational helicopter from the flight line. This helicopter underwent a few one-time modifications which had no impact on aircraft availability. The modified vehicle could then be wheeled into the training facility where three large screens provided out-the-window visual simulation. The visual system's wide field of view (FOV) provided adequate peripheral visual stimulation, enough that the absence of a motion base did not seem to be a critical limitation to training effectiveness.
Although the audio sub-systems employed are quite normal, the TRIAD approach was anything but ordinary. Not only did RSI expend considerable effort on frequency analysis of the entire audio spectrum, but they also studied the power levels associated with helicopter sound. By employing "a wall of amps and equalizers", the TRIAD recreates quite realistically the sensation of flight in the helicopter. Although actual savings are not readily attainable, certainly more than 50% of hardware costs were saved by utilizing the aircraft as the basis for the simulation.
Another interesting example was produced by the engineering staff at the Naval Air Test Center (NATC) in Patuxent, MD. One of the original F-18 aircraft, after having lived its useful life, was resurrected as an avionics analysis device. Modern aircraft have quite a complex internal avionics bus structure, and NATC decided to put the F-18's bus system to use for creating a cockpit simulator. To gain access to the aircraft bus, NATC literally cut a hole in the side of the building and pushed the nose of the aircraft into the lab. An interface was developed to connect the bus to the lab's mainframe computer, and - presto! - instant simulation. All the cockpit's multifunction displays, with their own internal symbol generators, required only stimulation from bus signals. With but a little engineering effort, "flying" NATC's static F-18 was providing pilots with the same cockpit presentation as produced by an F-18 when airborne.
Although several approaches have been developed to support motion cues for ground based simulation, none can faithfully reproduce sustained "G's". The only exception might be the Dynamic Flight Simulator (DFS) Facility at Naval Air Station, Warminster, PA. The DFS includes a 10' cockpit gondola on the end of a 50' boom which rotates at 48 RPM. It is quite capable of maintaining "G" forces to the pilot during his "flight." These "G" forces, experienced by pilots during combat maneuvers, must be endured on a regular basis for the pilot to develop an acceptable tolerance level. Several studies have resulted in the same conclusion - that frequent exposure to a high "G" environment is required to develop "G" tolerance.
A single aircraft can engage in high "G" maneuvering, but the cost per flight hour for today's fighter planes all but precludes such a practice for routine training purposes. And, introducing real bogey aircraft is a rarely afforded luxury. Although threat aircraft are essential to close combat maneuvering practice, simulated aircraft can certainly be used for radar intercept training.
Computer simulators can provide adequate threat simulation, replicating the on-board radar displays. Artificial Intelligence, or more accurately, an expert system may be employed to derive threat symbol reactions to pilot maneuvers. The military has already achieved cost reductions by employing a fleet of Lear jets to act as bogey aircraft. Imagine the even greater savings that could be realized by replacing the Lear jets with on-board microprocessors.
Future Prospects for Embedded Training
Aviators realize the actual aircraft cannot be completely replaced as a training medium, yet there are many situations where the simulator offers superior training. One of the most common applications of simulation on devices today is the practice of instrument flight in IFR conditions. Instrument Flight Rules apply during the periods of flight when the pilot does not have a discernable horizon and must therefore rely on his internal instruments to control aircraft altitude and to navigate.
Two additional obvious uses for simulators are emergency procedures training and tactical flight training. It is hardly feasible to practice such situations as an actual fire in flight, flight with a separated engine, or landing a jumbo jet with its wheels stuck in the up position. An operational flight simulator provides the perfect training vehicle for such scenarios. Using simulators to train for combat situations has been accepted as essential for many, many years. Military aviators in particular are not often permitted to expend actual munitions in training activities: the cost of some exceeds one million dollars per expended round. Simulation satisfies the training scenario perfectly, as the cost of a simulated missile is obviously zero. But further, what aviator, even a top gun instructor cares to be the target for the day in a live missile shoot? The newest air combat simulator systems allow as many as 12 "aircraft" to fly through sky, firing guns and missiles, keeping score of who shot who - all with no danger to pilot or aircraft.
To accentuate the potential for embedded training in aviation, it may help to more closely examine instrument procedures flight training. Until the advent of visual simulation, student pilots could only fly their aircraft to "minimums". Minimums refer to that period during an approach to landing that, when reached, if the pilot cannot visibly see the landing space area and maneuver his vehicle to a safe landing, he must execute a wave-off or aborted approach procedure. The pilot must add power and either circle the airport for another attempt to land, or at his discretion fly to an alternate airport. The pilot must fly his aircraft tot he exact point on the ground designated as the minimum approach point, all the while maintaining the aircraft at a proper airspeed, altitude, and attitude. Yet, even more demanding for the pilot is the mental and physiological challenges associated with transferring his thought process from the instrument world to a visual world.
During instrument approach, the pilot has a myriad of instruments depicting progress. For example, the airspeed indicator is obviously describing the aircraft's speed. But not so obvious is that the airspeed reading is not complete information. It provides the pilot with only his speed through the air mass without consideration to the speed of the air mass over the ground; the effect of any wind is not presented. Yet, the wind's effect is a critical variable. When flying at altitude and under cruise power, the wind has no effect on the safety of flight. The aircraft, except in the most extreme cases, maintains a higher speed than the wind speed, and the net effect is ground speed. This either hurries the plane to its destination, or, as is usually the case, the "wind is on the nose" delaying the aircraft's arrival.
Danger enters the picture, however, if the aircraft slows to a speed that is not all that much faster than the speed necessary to maintain safe flight. This normally occurs only just prior to landings and is known as the approach speed. Danger from unintentionally reaching or exceeding this minimum speed is especially prevalent during periods of high winds. High winds become dangerous when they are similar or greater in magnitude than the difference between minimum safe flying speed and approach speed. Winds that fit in this category are often associated with summer thunderstorms.
This problem is compounded because these types of winds are not steady, changing often in intensity and in direction as well. What may be at one minute a 40 knot head wind may suddenly become a tail wind, or vice versa. In either event, this can have a drastic effect on aircraft performance - in other words, on its ability to fly. This was one of the major contributions to a recent commercial airline crash. The wind played havoc with the air mass about aircraft, the aircraft stalled, and was "pushed" into the ground. Clearly even one of the most direct and informative instruments, the airspeed indicator, does not truly provide the complete picture about the parameter it serves to monitor.
All this discussion about the airspeed indicator, and this is only one instrument! The pilot must repeatedly examine several independent instruments to position his aircraft in the correct position in space.
The altimeter, as another example, would appear to present its information simply and completely. (We will discount variances between reported and actual barometric pressure as being insignificant, and also assume flat terrain - quite an assumption in itself!) The inadequacy of the altimeter's information is not in the determination of the current altitude, but in its ability to provide the correct altitude only when over the proper point on the ground. What advantage is it to be down to 400' if you are 1/2 mile closer to the touchdown point that planned?
The advent of ILS needles and course deviation indicators provides a dynamic snapshot of the aircraft position relative to desired glidepath and course. The glideslope logic indicator portrays height (e.g., a horizontal bar representing the proper altitude at any given time). If the bar is above the middle of the indicator, then the aircraft is below the glideslope and must climb up to the proper point. The correct procedure, rather than "fly up", is generally to reduce the rate of descent in order to intercept the glideslope. But, a question remains - just how low is the aircraft, and just how much should the descent rate be adjusted?
Answers to these questions, not provided by the altimeter, are derived from either the pilot's experience, or yet another instrument, the flight director. Although designs vary, the principle is the same. Usually superimposed over the primary altitude indicator (a gyro), are two orange tabs replicating wings that form a triangular slot. The pilot controls his aircraft and "flies" formation with the indicator. Obviously wisdom dictates that the pilot cross check primary flight instruments, but in theory, the aircraft could be piloted with reference only to the flight director.
In addition to glideslope information and descent rate direction, the flight director also gives commanded heading directions to the pilot. As with the glideslope indicator on the ILS indicator, the course deviation bar tells the pilot whether he is left or right of course. But again, how far off course? These two needles show angular displacement. Each needle width represents a proportional number of degrees deviation from the desired course. Now multiply that value times the miles (or is it feet?) from the touchdown point to determine the deviation from the centerline. While doing this, the pilot must diligently control attitude, altitude and airspeed. Yet, determining the deviation is not even the half of it!
A heading adjustment must be made. How much of a turn should be made from assigned heading; and how long until the return to the original heading? Or, should the pilot maintain a certain added heading to compensate for surface winds which are blowing the aircraft off course. Certainly the pilot has very much to do in those final seconds before landing. (Remember, the airplane is still in the clouds affording the pilot no visibility of the airport.) Assuming the descent into visual conditions occurs before mandatory waveoff, the pilot must identify the runway, line up his aircraft, and land. If his path properly followed the one portrayed by his instruments, the runway should be aligned directly in front of the airplane.
If the aircraft is not piloted accurately on the approach, the transition to visual flight becomes more difficult. There is a critical point to consider, and it is one that often confuses student aviators. Cockpit indications that the plane is left of course can induce the pilot to look to the right when breaking out below the clouds. A left-of-course situation is usually followed by a heading correction to the right. This results in the nose of the aircraft being pointed to the right of the intended touchdown point. As the pilot gets his first view of the runway, he is surprised to find that even though his aircraft is left of course, the runway is still on the left side of his field of view!
Now that your adrenalin is flowing and your fear of flying has escalated - don't be concerned. Even though the scenarios just described do provide a significant amount of work for the pilot, it is something that he is well trained to handle. The important point for this discussion is that there are several types of information displays available to the pilot, and the abiding question is, how much information is being presented in the most complete, most easily understood manner possible?
A book with hundreds of pages, and thousands of words, for example, certainly contains a wealth of information that is "transferred" to the reader. But the expression, "a picture is worth a thousand words" doesn't need explanation. Imagine, if you will, the application of the above cited quote. Imagine replacing (augmenting) the myriad of displays with an on-board graphics representation of the world in front of the aircraft. What pilot would not enjoy the opportunity to "peek through" the clouds while descending through 500' to assure himself that his aircraft is properly aligned with the runway? In effect, an on-board CGI system could provide the pilot with a constant window through the weather.
All raw material necessary to accomplish such instrumentation is already in place. The same ground-based electronics that transmits the information to the aircraft to position course deviation indicators could be fed directly to an on-board CGI system. The CGI system would then be loaded with the data base representing the destination airport. At any time the pilot wishes, the CGI "window through the weather" can be called upon to present a visual representation of the airport environment exactly as if the aircraft were approaching to land in visual flight conditions. Imagine, if you will, that whatever the weather - fog, clouds, rain, snow, even the darkest night - the circumstances can be transformed into day VFR landing conditions from the pilot's point of view.
Science fiction? A scenario that is not obtainable with today's technology? No. In fact, CGI systems are already in production that consume less volume and less weight than existing avionics equipment. Any number of today's aircraft could be retrofitted with such a system to provide not only a perpetual VFR view of the airport to the pilot, but any other type visual cues as well. Such embedded CIG systems could allow pilots to practice landings at airports without descending below an arbitrarily selected altitude; military pilots could practice maneuvers in simulated environments, feeling the real stress of flight but without the associated dangers of flying near the ground.
The clear and fascinating conclusion is that the same devices now categorized as tools for embedded training will most probably become the elemental tools of flight in the months and years to come.