Optimal complementary filters and their application in motion measurement

Abstract

A variety of sensor types are normally available for measuring the same physical quantity, and each type exhibits its own error characteristics. Thus the best measurement can sometimes be obtained by combining information from two (or more) sensors. This paper proposes an optimal H∞ method to design complementary filters for combining two sensor signals. The method relies on selecting weighting functions which represent the measurement noise or other error characteristics of the sensors. Two detailed industrial examples are described, concerning the measurement of position and velocity for high-performance testing machines. Experimental results are presented, and the advantage of the proposed design method compared with heuristic approaches is demonstrated.