Fault-tolerant attitude computation for unmanned aerial vehicles

Abstract

Attitude information is essential in the control of unmanned aerial vehicles (UAVs). One way of defining attitude is through Euler angles. These angles can be determined based on the measurements of the projections of the gravity and earth magnetic fields on the three body axes of the vehicle. Twenty-five methods have been developed to compute the Euler angles and each of these methods employ a subset of the six measurements. The capability of computing the Euler angles in multiple ways provides a diversified redundancy required for fault tolerance. The proposed approach can identify the sensor failures and even separate the reference fields from the disturbances. A bank-to-turn manoeuvre of the NASA GTM UAV is used to show how to determine the correct Euler angles despite interferences by inertial acceleration.