Answers are grounded in this paper and cite their sources.
State estimation becomes challenging when measurements are subjected to unknown saturation thresholds. With unknown thresholds, the traditional approaches, such as the Extended Kalman Filter and the Tobit Kalman Filter, become unreliable. This limitation is demonstrated through a performance analysis of these estimators. To address this issue, an adaptive Unscented Kalman Filter that jointly estimates the system states and the unknown saturation thresholds is proposed. In the proposed framework, sigma points are generated under the assumption of Gaussian state distributions, while the predicted measurements are clipped according to the estimated thresholds. The prediction is then refined in the update step, where the process noise covariance is adaptively adjusted based on a geometric moving average of the normalized innovation squared. The simulation and the hardware experiment confirm that our framework provides accurate estimates of both states and thresholds that closely capture the true values.
No takes yet. Share an insight, caveat, or question.
Jetawatthana et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: