Learning-Based Fault-Tolerant Control for an Hexarotor With Model Uncertainty

Abstract

In this paper, we present a learning-based trackingcontroller based on Gaussian processes (GP) for a fault-toleranthexarotor in a recovery maneuver. In particular, we use GPsto estimate certain uncertainties that appear in a hexacoptervehicle with the ability to reconfigure its rotors to compensatefor failures. The rotors reconfiguration introduces disturbancesthat make the dynamic model of the vehicle differ from thenominal model. The control algorithm is designed to learn andcompensate for the amount of modeling uncertainties after afailure in the control allocation reconfiguration by using GP asa learning-based model for the predictions. In particular, thepresented approach guarantees a probabilistic bounded trackingerror with high probability. The performance of the learning-based fault-tolerant controller is evaluated by experimental testswith a hexarotor UAV.