Flight Evaluation of Adaptive High-Bandwidth Control Methods for Unmanned Helicopters

Abstract

This paper summarizes the formulation of two distinct methods for high-bandwidth nonlinear adaptive flight control of an unmanned helicopter testbed, and presents an evaluation and comparison of the two controllers in both simulation and flight test when subject to unmodeled dynamics, time delay, and actuator position and rate saturation. The attitude command system functions as an inner loop, and features a dynamic inversion control law augmented by an on-line neural network cast in an output feedback setting. The inverting control law is based on a simple linear model of the helicopter and its actuators at the hover condition. Both a direct and an observer-based method for output feedback design are evaluated. Pseudo control hedging is employed to prevent adaptation to input characteristics such actuator position and rate saturation. A previously developed adaptive trajectory controller forms an outer loop to the attitude command system. A brief description is given of the unmanned helicopter testbed, real-time control system implementation, simulation, and flight test facilities. Stable adaptation and a design natural frequency for the inner loops of up to 10 radians per second is demonstrated in the presence of actuator and control rotor dynamics, system time delays, sensor noise, sensor latency, gusts and input saturation.