NASA Logo, National Aeronautics and Space Administration

+NASA Home

+Ames Home

Learn To Fly Tests to Take Place August 14-25 With Ames-Developed Optimal Control Allocater
Intelligent Systems Division Banner

Learn To Fly Tests to Take Place August 14-25 With Ames-Developed Optimal Control Allocater

The Convergent Aeronautics Solutions’ (CAS) Learn to Fly (L2F) project will conduct two weeks of flight tests of the Woodstock 1 glider and E1 electric airplane at Fort A.P. Hill near Bowling Green, Virginia, from August 14-25. The aircraft will be deployed from a balloon system during multiple test flights to validate real-time system identification, modeling, learning, and control algorithms. After being dropped from the balloon, the aircraft will autonomously fly to several waypoints and then land. A remote control pilot will be on standby to take over control of the aircraft in the case of unexpected behavior. The flight control system will have a very low-fidelity model of the aircraft when it is first launched from the balloon. After launch, the real-time system identification and modeling will construct aerodynamic models during flight that are sent to the flight control system to enable the aircraft to be autonomously piloted along the predefined flight path. The flight control system will learn and adapt to its parameters during flight to achieve robust control of the aircraft. Ames L2F team members, Susan Frost and Wendy Okolo, will be on standby during flight testing to remotely support any needs related to the Optimal Control Allocator (OCA) developed for the aircraft.

BACKGROUND: The current transition between aerospace vehicle design and the development of an operational, robust, aerospace vehicle is a sequential, iterative, labor and cost-intensive process that can be replaced with real-time methods applied in flight. The L2F project is a paradigm change on this process that merges real-time vehicle aerodynamic modeling with learning/adaptive control developed by researchers at the Armstrong Flight Research Center (AFRC), Ames Research Center (ARC), and Langley Research Center (LaRC) on a novel unmanned aircraft configuration. L2F enables an aircraft to learn its aerodynamic properties and adjust its control parameters during flight using advanced real-time system identification, modeling, and adaptive control algorithms. Since the algorithms work in real-time, an aircraft can be flown with only a rudimentary model of its parameters, thereby reducing the total time spent developing an operational aircraft.

LaRC researchers developed the real-time system identification and modeling software and the learning and adaptive flight control algorithms. AFRC provided support for flight tests. The ARC team developed the Optimal Control Allocator (OCA) for the L2F flight control system. The simulation architecture design is modular, supporting experimentation with different OCA algorithms and technologies. Using frequently updated information from the aerodynamic modeling, the control allocator determines the optimal control surface positions to achieve the body axis moments (roll, pitch, and yaw) commanded by the flight control system while minimizing drag due to control surface deflection. The control allocator, also called a mixer, uses an active set method to solve the weighted, bounded least-squares problem that will achieve the desired moments and drive the control surfaces to their preferred positions. The OCA, which can be utilized on any over-actuated vehicle (more control surfaces than moments), was chosen for its numerical and convergent properties, and its ability to evenly distribute the control effort among the aircraft’s control surfaces.

NASA PROGRAM FUNDING: Convergent Aeronautics Solutions (CAS), Aeronautics Research Mission Directorate (ARMD)


ARC TEAM: Susan Frost, Wendy A. Okolo, Richard Sayre, and Christopher Teubert


First Gov logo
NASA Logo -