NASA recently awarded University of Illinois MechSE Professor Naira Hovakimyan $1.5 million to develop an integrated reconfigurable controller for vehicle resilience (iReCoVeR) that will enhance next-generation aviation safety. The iReCoVeR architecture is based on Hovakimyan’s L1 adaptive control methodology, which is a promising technology for loss-of-control situations—the leading cause of commercial airline fatalities during the last 20 years.
Loss of control refers to situations in which the aircraft is flying outside its normal flight envelope, or the margins within which the aircraft is structurally designed to fly safely. Ice forming on the wings, wake vortices, wind gusts, and bird strikes could all cause loss of control, which makes it difficult for the pilot to stabilize the aircraft.
With the NASA grant, Hovakimyan and her research team will integrate the L1 adaptive flight controller with flight envelope estimation and protection schemes, an upset onset detection scheme, and a fault detection and isolation module.
“The iReCoVeR system will be designed to perform under a combination of pilot input errors and multiple adverse conditions, including sensor and actuator failures, vehicle impairment, turbulence, and wake vortices,” said Hovakimyan, noting that her team will place special emphasis on sensor failures and adverse icing conditions.
In the past, adaptive control schemes could only guarantee stability in steady-state operations; they could not adapt to unexpected circumstances. With L1 architectures, the controller maintains robustness, which enables the system to keep functioning—with a priori predictable performance—despite rapidly changing operating conditions.
L1 is able to compensate for large changes in system dynamics, which is important for providing consistent pilot handling qualities in off-nominal adverse conditions. Another benefit is that it keeps the plane within the boundaries of the flight envelope for a few more seconds than existing control technology, thus giving the pilot additional valuable time to regain control. Moreover, L1 may help reduce the need for gain-scheduling, a common non-linear control process that is expensive and time-consuming because it requires designing a different control for every flight condition.
According to NASA Senior Research Engineer Irene Gregory, L1’s biggest advantage over other control methods is its ability to deal with uncertain, rapidly changing dynamics in a predictable way. “The design process is very systemic and follows the long-established practice of trading off performance and robustness in a conventional way,” said Gregory.
Illinois aerospace engineering graduate students Enric Xargay and Ronald Choe are working with Hovakimyan to add new features to the L1 adaptive control technology, including flight envelope protection schemes for preventing loss of control. These new features may prevent crashes like the 2001 disaster near New York City. In that instance, an American Airlines jet flew into the wake of another plane and rolled on its side. In trying to stabilize the plane, the pilot applied too much rudder causing the plane’s tail to break, which resulted in a fatal crash.
“We’ll create a system that recognizes if the pilot is being too aggressive,” said Xargay who is developing the algorithms to recognize whether a pilot’s corrective actions during an emergency are suitable and capable of returning the plane to stable flight conditions. “Our system would take the good part of what the pilot is doing, but leave his bad actions out.”
The NASA grant is a continuation of work that Hovakimyan started several years ago and culminated with NASA Langley adopting L1 as the baseline control law for the next-generation of flight vehicles at NASA’s AirSTAR test facility, which uses sub-scale planes to test loss of control technologies. IEEE Control Systems Magazine featured L1’s NASA flight test results in its October 2011 issue on safety-critical control.
“Being a baseline control law means L1 adaptive control will function as the default control law that the pilot will be flying to take the aircraft into adverse conditions for any particular research objectives and to safely recover from these conditions,” said Gregory.
Hovakimyan is collaborating on iReCoVeR with University of Illinois Computer Science Professor Alex Kirlik and Aerospace Engineering Professor Mike Bragg, as well as University of Connecticut Professor Chengyu Cao, who co-developed the L1 adaptive control theory starting in 2004. She will validate the iReCoVeR architecture on a full-scale aircraft model flight simulator at the University of Illinois Beckman Institute with a pilot in the loop, before transitioning testing to NASA facilities.
Aviation isn’t the only application for Hovakimyan’s L1 adaptive control method, which industry is adopting for energy production and national security. Norwegian energy company Statoil is applying L1 to off-shore oil and gas production and exploration. In October, Statoil sent Hessam Mahdianfar, a doctoral student from the Norwegian University of Science and Technology, to Hovakimyan’s lab to learn L1. “We saw L1 was very successful in aerospace engineering, so we thought we should try it in the drilling industry, too,” said Mahdianfar, who is learning how to apply L1 to managed pressure drilling (MPD), a relatively new technology to extract oil reserves from deep-sea fields.
MPD is a promising technique because it allows drilling in deep, high pressure and high temperature environments. “The industry has been looking at L1 for about 3-4 years because the situation on deep, offshore oil wells is really uncertain,” said Mahdianfar. “We think the L1 controller might be useful in practice to guarantee a safe and secure drilling system.”
As for national security applications, Hovakimyan is collaborating with the Naval Post Graduate School on a project to use L1 technologies to improve the robustness of fleets of drones performing intelligence, surveillance, and reconnaissance missions. To date, the Naval Postgraduate School has satisfactorily performed more than 100 flights with L1 flight control systems onboard unmanned aircraft. These flight tests were featured in the October 2012 issue of IEEE Control Systems Magazine.
Hovakimyan was pleasantly surprised recently to learn about the success a Danish research team has had with L1 by implementing the technology in an unmanned watercraft it developed for the Danish Royal Navy to use in security and search and rescue operations. The team received a prestigious best paper award at an IFAC workshop for their work.
“It’s a success story, of which we had no idea,” said Hovakimyan. “The Danish team’s work was based on our book and it shows just how far the L1 adaptive control theory has transitioned. L1 was developed enough for them to [successfully] use it without even contacting us.”
According to lead researcher Casper Svendsen, the Danish team deployed L1 in its design of the jet ski autopilot because L1 enables the controller to adapt to changing dynamics of the vehicle while also ensuring stability. In other words, as the vehicle accelerates on a choppy sea, it will function properly. “The L1 adaptive control theory has shown to guarantee both robustness and performance of the closed loop system in different operational conditions,” said Svendsen.
Internationally, Hovakimyan’s work is earning praises. In 2012, she received a Technical Achievement Award from the International Conference on Nonlinear Problems in Aviation and Aerospace, and in 2011, the American Institute of Aeronautics and Astronautics (AIAA) presented her with its Mechanics and Control of Flight award.