Online Sensor Bias Estimation & Calibration by Kalman Filtering with Adaptive Lyapunov Redesign Method

Faulty measurements bring faulty controls and the major part of those are the bias in sensor readings. In this paper, the theory of adaptive and online calibration of a sensor with another measurement model is demonstrated by using Lyapunov redesign approach in Kalman filtering framework. In the measurement update step of Kalman filter, the effect of measurement bias is considered as a parametric uncertainty term and it is proven that it can be regressed by a universal approximator and can be eliminated from measurement update step while preserving the asymptotic stability of the estimator. Then, the convergence criteria for online parameter adaptation are obtained. Finally, a case study for estimation-based roll control of a missile is conducted and the results of online calibration is discussed.