Envelope Protection for Autonomous Unmanned Aerial Vehicles

This paper describes the design, development, and testing of an automatic envelope protection system as implemented on Georgia Institute of Technology's unmanned helicopter GTMax. The envelope protection system makes use of online-learning adaptive neural networks to generate online dynamic models, which are used to estimate limits on controller commands. The system provides command capability up to the limit boundaries while preventing envelope exceedance. Simulation and flight-test results are provided for load factor and rotor stall limit protection during aggressive maneuvering.