Ground Optical Signal Processing Architecture for Contributing Space-Based SSA Sensor Data.

摘要

DARPA's OrbitOutlook aims to augment the performance of the Space Surveillance Network s Space Situational Awareness (SSA) by adding more data more often from more diverse sources to increase Space Domain Awareness (SDA) and determine when satellites are at risk. OrbitOutlook also seeks to demonstrate the ability to rapidly include new instruments to alert for indications and warnings of space events. [1] In order to accomplish this, an underlying framework to process sensor data and provide useful products, such as angles-only measurements, is required. While existing optical signal processing implementations are capable of converting raw sensor data to angles-only measurements, they may be difficult to customize, obtain, and deploy on low-cost demonstration programs. The Ground Optical Signal Processing Architecture (GOSPA) (inexpensive, easy to customize, obtain, and deploy) can ingest raw imagery and telemetry from a space-based optical sensor and perform a background subtraction process to remove anomalous pixels, interpolate over bad pixels, and dominant temporal noise. Streak end points and target centroids are located using the AFRL Generalized Electro-Optical Detection, Tracking, ID, and Characterization Application (GEODETICA); an image data reduction pipeline for ground and space-based SSA. These identified streak locations are fused with the corresponding telemetry to determine the Right Ascension and Declination (RA/DEC) measurements. The measurements are processed through the AFRL Constrained Admissible Region Multiple Hypothesis Filter (CAR-MHF) to determine its initial orbit (IOD). GOSPA performance is demonstrated using non-rate tracking collections against a satellite in GEO, simulated from a visible optical imaging sensor in a polar LEO. Stars, noise and bad pixels are simulated based on look angles and sensor parameters.