Fault and Failure Tolerant Model Predictive Control of Quadrotor UAV

This paper presents fault tolerance control (FTC) of a quadrotor under actuator faults. A complete FTC design approach with fault detection and diagnosis (FDD) is addressed. The proposed FTC is based on the model predictive control, which can be applied to nonlinear systems using the so-called successive convexification algorithm, which converts a nonconvex function to a convex function in a successive manner through linearization. The faults are parameterized in the form of loss of effectiveness (LOE) of a quadrotor actuator and estimated using a two-stage Kalman filter. Compared with other FTC, the proposed FTC is capable of controlling not only partial loss of actuator effectiveness defined as a fault but also a complete loss of actuator effectiveness defined as failure. Controller switching is not required, while an actuator saturation limit is also considered. The proposed FTC approach has been validated under various actuator fault conditions through nonlinear simulation studies.