Sensor Fusion with Censoring Limits
Bethany L. Allik · AIAA Scitech 2020 Forum · 2020
Sensor fusion enables higher precision and wider sensing ranges when only multiple low quality sensors with limited dynamic range are available or afforded. Here, the sensor fusion approach is used for the radar tracking problem to extend the range of tracking capability. The proposed approach is performed with specific consideration to censoring limitations of a radar system. In this paper a Cram\'er Rao lower bound for a censored measurement model is derived. Motivated by the results of this bound the Tobit Kalman filter is used as a mechanism for data fusion amongst sensors. An event based system is proposed, where the probability of censoring will dictate whether specific measurements are used. This approach can be used for stacked sensors over large measurement regions, and, for applications with measurement redundancy. The method is demonstrated with a radar tracking problem, and shows great tracking ability of the state, even when the latent measurement is not being observed for some time.