Adaptive Control Inversion and Actuator Failure Compensation for UAVs

Abstract

For MIMO systems where it is desired to control n state rates with n actuators, this paper proposes a method to: 1) invert the design system’s control dynamics to construct a control signal, 2) simultaneously estimate and adapt to differences between real and design actuator effectiveness, and 3) automatically switch control goals to attempt to maintain critical control upon detection of (an) ineffective actuator(s). The estimation and adaptation is based on a model reference adaptive control (MRAC) scheme similar to [1].