A sensor fusion approach for exploiting new measurements in an existing controller

This paper considers the problem of improving an original controller once a number of new measurements become available. We consider the situation, where for commissioning reasons or other reasons, it is required that the original controller stays in place (e.g. in case it is integrated in the actuator). Therefore, the new measurements can be used only to improve the inputs of the original controller. To that end, a sensor fusion approach is taken. An observer based architecture is proposed. For this observer based approach, a separation theorem is proven, which under certain necessary and sufficient conditions specialize to a double separation principle. The main novelty in the paper is the incremental approach, which is obtained by combining a couple of standard techniques. A numerical example is given, which illustrates the potentials for improvements of performance.