Adaptive control of the ball and beam plant in the presence of sensor measure outliers

The adaptive control of plants whose sensor measurements are corrupted by outliers is considered. Outliers are large deviations of the signal being measured, only occurring in a few percent of the observations. Although rare, due to their large amplitude, outliers heavily contaminate the output of noise removing filters whose design rely on the minimization of a quadratic loss. If an adaptive controller is used, its performance in the presence of sensor measurement outliers degrades. This problem is tackled in this paper. Methods for outlier removal which are suitable for controller applications are reviewed. Their incorporation with advantage in an adaptive control loop is illustrated through an application to position control in the ball and beam plant.