Kalman filtering with indirect noise measurements

Abstract

Accurate remote tracking of fixed objects from a moving platform requires overcoming platform noise. Such tracking becomes difficult when the only inertial reference (such as a central aircraft inertial navigation system) is remote from the sensor, which experiences non-measured angular movements due to airframe vibrations and flexing. In such a scenario, Kalman filtering cannot converge on a true value because all noise sources are not known. Current naïve approaches arbitrarily boost noise with fixed additive or multiplicative factors. However, such approaches slow filter response and; thus, often fail to give timely results. Embodiments of the present disclosure derive inertial reference parameters to quantify noise of the sensor that is remote from the inertial reference. Advantageously, disclosed embodiments enable use of remote sensors with an existing inertial reference, rather than consolidating sensors and the inertial reference at a single location or providing inertial references at each sensor.