Flight testing of a reconfigurable control system on an unmanned aircraft

Abstract

A radio-controlled aircraft was built and equipped with air-data and inertial sensors. A radio frequency link was added to transmit data and receive commands from a ground station. Data from several flight tests were used to characterize the dynamic response of the aircraft. Despite the high level of noise associated with the low-cost sensor suite, consistent identification of critical aircraft parameters was obtained. Flight tests were also conducted with actuator failures induced on one elevator, one aileron, and one engine. Recursive parameter identification produced parameters tracking the effects of the failures, such as reduced effectiveness of pitch commands due to a locked elevator, or roll and sideslip due to engine failure. The identified parameters were also used in reconfigurable control experiments, where knowledge of the aircraft parameters was used to compensate for the effect of failures, reducing the pilot's workload. Overall, the paper demonstrates that recursive identification and reconfigurable control algorithms are implementable in real-time, even in low-cost platforms. They can be designed to effectively compensate for actuator failures and aircraft damage.