An analytical fault diagnosis method for yaw estimation of quadrotors

A new analytical fault diagnosis method is presented for the yaw and yaw rate estimation of a quadrotor. In the existing methods, the faults of the inertial sensors and the measurement sensors are treated separately. The presented method can deal with the faults of both the z-axis gyro and the magnetic sensor by introducing the torque model, which is used as a virtual sensor. The filter design, fault detection, isolation and recovery strategies are discussed. Real flight data is used to validate the proposed approach, showing that both the z-axis gyro fault and the magnetic sensor fault can be detected.