Electronic feedback control systems have been employed for many decades with great success to stabilize a wide variety of piloted aircraft, and to improve their handling characteristics. This is also true of remotely piloted aircraft, and more recently fully automated, or “autonomous” flight systems. A commanded signal (for example, pitch attitude) is subtracted from a measurement of that quantity to obtain an error signal. The error is then multiplied by a gain to produce a control signal that when added to the nominal pilot command improves the aircraft’s dynamic response (e.g. made stable, and/or performance in tracking the command improved).
As control engineers, we often employ a model of the aircraft dynamics, linearized about a given flight condition (e.g. hover), to design gains for the feedback loops. A wide variety of very mature linear design tools exist to ensure the designed system will be stable and well behaved. To apply this design method in practice, we must recognize that aircraft dynamics are nonlinear, and we thus schedule the gains obtained from a number of individual point designs across the flight envelop with variables such as airspeed. This technique has been used with great success in a wide variety or programs over the years, but incurs one very large burden. It is necessary for success to either (1) spend a significant amount of time and money to develop an accurate model of the vehicle dynamics across the flight envelop to support the design process, or (2) spend a significant amount of time and money to experimentally tune the gains of the controller across the flight envelop. In many cases, some combination of the two is employed. In either case, even a small change in the vehicle configuration can result in a requirement to revisit the control system design, and to revalidate it in flight test. This change can occur because of a requirement to alter the system during design, or due to a desire to alter the system in production to accommodate a new payload system or other improvement. Furthermore, changes can occur in flight due to a fault or failure, or due to long-term evolution of the system to a new variant or model. Failure to recognize changes in the vehicle dynamics in the control system can result in poor control system performance or instability.
To overcome this burden, Guided Systems, in partnership with the Georgia Institute of Technology, and with support from the U.S. Department of Defense (DoD), has developed, validated and transitioned a new adaptive control technology tailored to the requirements of aircraft applications. This technology largely eliminates dependence of the control system design on modeling, and when supported by physical redundancy in control actuation, creates tolerance in flight of gross changes in dynamics due to faults, failures, or battle damage. Guided Systems has developed a broad patent portfolio in this technology area, and completed a number of DoD-funded flight demonstrations to establish the credibility of the technology.
For example, in a United States Air Force (USAF) program entitled “Reconfigurable Control for Tailless Fighter Aircraft” or RESTORE, our adaptive technology was applied in partnership with Boeing to manage the stability and handling qualities of a remotely piloted scale model of an advanced tailless fighter design that employed thrust vectoring. Adaptation was used to overcome a wide variety of simulated faults, failures and battle damage, first in simulation, and ultimately in flight test of the X-36 technology demonstration aircraft. This program clearly illustrated the fact that even when sufficient financial resources are available for wind tunnel testing and flight validation of dynamic models to support control system design, these models may no longer be valid once the system changes due to a fault or failure. Adaptation was shown to be very effective in maintaining stability and performance in such conditions.
In another USAF program, Guided Systems lead a team that included Georgia Tech and Boeing to design, build and flight demonstrate an adaptive autopilot for application to multiple variants of the Joint Direct Attack Munitions (JDAM). The original JDAM program produced a kit with Global Positioning System (GPS) receiver, inertial sensors, navigation computer, and actuated tail fins that can be fitted to pre-existing bombs by replacing the tail section and thus produce a precision guided munitions. The original autopilot was developed using a gain-scheduled linear control law as described above, and as a result, each munitions variant had to be tested in the wind tunnel. Extensive effort was expended to validate the models and the controller in flight test, at great expense in both time and money. The adaptive autopilot developed by Guided Systems largely eliminates dependence of the control system design on high fidelity modeling, and allows a single autopilot design to be employed on multiple variants of JDAM (MK-84, BLU-109, MK-82, etc.). The program culminated in multiple flight tests of the MK-84, dropped from an F-16 at Eglin Air Force Base, and proved the viability of the concept and the value of the technology. Boeing has now incorporated this technology in its production of the JDAM weapon system.
As described in a later section, this advanced adaptive control technology is now available in a standard product for control of unmanned air vehicles (UAVs), and offers the potential for huge savings in both the cost of development, application and long-term maintenance of such control systems over the life of any given UAV program.