Every day, global sensor networks produce thousands of new, uncorrelated tracks (UCOs). Manually associating these with known catalog objects is a slow, error-prone process performed by analysts, creating a latency that directly increases collision risk. This catalog drift—where objects are misidentified or remain unknown—degrades the accuracy of all downstream conjunction assessments. Automating this correlation is not an AI experiment; it is an operational necessity to achieve the data fidelity required for safe, scalable space operations.




