Adapts To Changing Aircraft Conditions, Implications For Hypersonic Flight

An L1 adaptive flight control system flew on a NASA GTM aircraft (T2) on March 24, during the last deployment of the AirSTAR team at Fort Pickett, VA.  The flight, number 14 for the aircraft, lasted approximately 16 minutes, with the adaptive controller closing the inner-loop for about 14 minutes.
 
The first successful flight of the L1 adaptive controller on NASA's GTM is the result of work by Professor Naira Hovakimyan and PhD student Enric Xargay at the University of Illinois at Urbana-Champaign, and Professor Chengyu Cao, currently at University of Connecticut. The development of this theory started five years ago, when Hovakimyan and Cao were affiliated with Virginia Tech. The L1 adaptive control architecture implemented onboard is an all-adaptive control law, thus providing both nominal performance and mitigating performance degradation due to faults and failures.
 
"The architectures of L1 adaptive control theory have guaranteed transient performance and guaranteed robustness in the presence of fast adaptation, without enforcing persistence of excitation, without any gain scheduling in the controller parameters, and  without resorting to high-gain feedback," explained Hovakimyan, who is a Schaller Faculty Scholar and professor of mechanical science and engineering at Illinois. Moreover, the current L1 architecture accommodates uncertainty that is not directly controllable (unmatched) and cannot be addressed by recursive design methods, like backstepping.

NAS Adaptive Control Test Aircraft

"It was the first flight of a direct all-adaptive controller with a pilot in the loop," according to Dr. Irene Gregory, the project's point-of-contact at NASA. The T2 is a twin-engine jet-powered and dynamically-scaled (5.5%) civil transport aircraft, designed and instrumented to perform control law evaluation, experiment design and modeling research, in-flight failure emulation, and flight into upset conditions. This is the first successful flight of an all-adaptive flight control system that deals with aircraft stability degradation and control surface failures.
 
"During 'Flight 14', both the longitudinal and the lateral stability characteristics of the aircraft were simultaneously degraded from nominal to 125% (a highly unstable open-loop aircraft)," explained Xargay, a graduate student in Hovakimyan's research group. "The version of the L1 controller that flew maintained nominal performance for both the 50% and 75% stability reductions scenarios.
 
"The pilot noticed a small amount of performance degradation was with 100% stability reduction," Xargay added. "When vehicle stability was reduced 125%, the L1 controller began to show closed-loop instability, as expected from prior piloted evaluations in NASA LaRC's simulator. The last flight card also considered the case of simultaneous locked-in-place failure of the two left elevators. The adaptive controller was able to guarantee safe operation of the vehicle and the pilot did not notice a change in aircraft behavior."

NASA Adaptive Control Aircraft

"Accommodating this class of uncertainty as well as fast and robust adaptation are some of the key features distinguishing the L1 adaptive control from other adaptive controllers that have flown in the past." Gregory said. "The flights of an all-adaptive control law significantly advance the current state of the art and enhance our understanding of the capability of the adaptive control as a class of controllers. While this first flight can be viewed as a big leap in the development of adaptive flight control systems, it only represents a small step forward towards airworthiness certification of these systems." Further development is underway and more challenging flight tests are planned for June 2010.
 
An adaptive control law was earlier implemented as an augmentation loop, so that when the vehicle was subjected to faults and failures, the augmented system maintained specified performance that is provided by a baseline controller tuned for flying qualities. In the case of an all-adaptive control law there is no baseline controller, and the adaptive controller provides both the nominal flying qualities performance and recovers that performance under faults and failures. This was the first flight test of a direct robust all-adaptive controller with a pilot in the loop.
 
Hovakimyan's  and Cao's new approach to adaptive control makes it possible to achieve fast adaptation with guaranteed performance. The L1 adaptive control theory guarantees the limits of stability and certain performance metrics that will ultimately reduce the cost of designing feedback control systems by enabling them to be validated and verified without months of expensive testing.

In addition to NASA, L1 adaptive control theory has attracted an impressive list of government and corporate sponsors that includes Boeing and all three branches of the U.S. military. Naira's research group is working with such partners to develop micro unmanned air vehicles, an autopilot design for aerial refueling, hypersonic vehicles (which could be used as a low-cost solution to space exploration), and other cutting-edge aerospace technologies.
 
"Adaptive control is flight critical," explained Kevin Wise, senior technical fellow of advanced flight controls at the Boeing Company in St. Louis, Missouri. "If there's a problem with the algorithm, it will cause a crash. Naira's theories have greatly improved adaptive control by improving the performance and reliability of algorithms. This will allow us to get the benefits of adaptive control without the risks of oscillation or instability." 

Completed T2 Airframe NASA Photo

Wise first met Hovakimyan while she was a research scientist at Georgia Institute of Technology. Later, when she was a faculty member at Virginia Institute of Technology, Wise told her about the problems they were having with oscillations in adaptive controllers, and her group addressed those. "Adaptive Control was a relatively new area for her, but she immediately understood what the problems were and took it to the next level," Wise said. "She has brought a fresh perspective to this type of work that has advanced the entire field."
 
Wise believes that the future of adaptive control for manned flight and hypersonic flight is still about 10 years away, yet Hovakimyan's work has already begun to pave the way.  "Researchers have already taken her algorithms and flown with them on small research planes," he said. "It takes time for an innovation of this type to make it into industry, but Naira's theories are helping to speed that transition."

FMI: www.nasa.gov