Kalman Filter with Complementary Constraint and Integrated Navigation Systems Applications

摘要

The report presents a method of navigation system integration that yields an optimal system regardless of the particular redundancy in available sensor data. The proposed mechanization is based on the criterion that the statistics of the signal are unknown. This results in the concept of distortionless or complementary filtering; an extensive review of this concept is presented. When the measurement errors for the Kalman filter are independent, it is shown that the inputs can be processed sequentially, even with the complementary constraint imposed. Thus, if there is an input that is not available, it is simply passed over, and an optimal estimate is still obtained from the remaining inputs. This even allows for a failure in the inertial navigation unit itself, whereas in most hybrid inertial systems to date this would be impossible. (Author)