SENSOR BIAS ESTIMATION AND UNCERTAINTY QUANTIFICATION STRATEGIES FOR SPACE OBJECT TRACKING

Measurements from any given sensor are corrupted by noise and are biased. The problems of estimation and uncertainty quantification of sensor biases are investigated in this paper. Several approaches to these problems are explored, and their success in the mitigation of bias is investigated. Filtering without compensating for bias, augmented filtering, and consider filtering approaches are studied and their results are compared. In addition, smoothing results are presented. The approaches have their own merits and drawbacks, and their pros and cons are discussed within and recommendations are made as to when to use which approach. Statistical consistency checks are provided to show when the filter is not performing as desired. The focus of this paper is on estimating biases that are assumed to be constant, but biases with a time varying structure can be accommodated if a sampling rate higher than the Nyquist frequency is available.