Sean Stratton, SciTec
Keywords: Data Fusion, RSO Tracking, Tracking Performance, RTN Reference Frame
Abstract:
Multi-sensor, multi-target track data fusion applications are designed to form time series state vector estimates (tracks) for a given stream of asynchronous measurements (observations). These data may come from a broad array of providers using independent instrumentation and processing methodology. In order for this information to be useable by the tracking application, vendors must adhere to a well-documented standard structure that captures the essence of the information being conveyed. For optical measurements, it is standard practice to define observations in terms of angles in a spherical polar coordinate system. While this choice is intuitive, this is a non-linear system and these angles are often treated as linearly independent variables. Consequentially, metrics that should only depend on relative differences between measurement and prediction, such as those used for measurement to track association and state vector update, instead have additional dependence on their absolute location within the coordinate system. There are also less impactful consequences such as visualization of tracks and errors where, even when proper corrections are made, lead to artifacts that can either trick the analyst into believing features are present in data that are not, or creates noise that detracts from real phenomena. SciTec proposes to address this problem by adopting the “Radial-Transverse-Normal” (RTN) reference frame as the vector basis for representing optical measurements. This provides the means to convey the same information while preserving the linear relationship between vector spaces when computing residuals between measurement and prediction.
In this presentation, SciTec will demonstrate the utility of defining optical measurements within the RTN reference frame over conventional spherical polar representation in the context of three common use cases. First, for anomaly detection, the choice of RTN coordinates eliminates confusing artifacts caused by the representation of residuals in a polar coordinate system. Second, for measurement-to-track association, RTN provides a basis that allows for the scalar Mahalanobis Distance metric to have uniform behavior across the celestial sphere. And third, for state estimation, measurements in RTN provide greater stability by preserving the linear relationship between measurement and state space.
Each case will be demonstrated using real-world example data provided under DARPA’s Space-domain Wide Area Tracking and Characterization (Space-WATCH) program.
Date of Conference: September 16-19, 2025
Track: Astrodynamics