A Tobit traceless Kalman filter technique TUKF: handle truncated data
Bo Su, Qingyue Yang, Bo Bai, Zeshan Yan, Lei Zhu, Shanliangkun He · 2021 IEEE 21st International Conference on Software Quality, Reliability and Security Companion (QRS-C) · 2021
We propose a Kalman filtering technique that handle truncated data, called Tobit Unscented Kalman Filter (TUKF). Based on description of truncated data, the new TUKF algorithm makes three major corrections to the traceless Kalman filtering algorithm, namely, the correction of the observation equation, the correction of the statistical properties of the measured observations, and the Implemented an extension of the TKF algorithm to UKF algorithm capable of handling. The new TUKF algorithm performs well in the data truncation problem and still maintains good target tracking capability in the detection range-constrained target tracking problem. In addition, the method can be applied not only to linear systems but also to more complex systems such as nonlinear systems.