Digantara Unveils MOSAIC Orbital Tracking Network

Each initial node combines an Optical Head Unit for precision sky imaging with an Electronics Head Unit for onboard processing, timing, communications, and autonomous operations, according to India Today. The Print reported that the nodes are designed to operate independently on solar power with battery backup and to withstand conditions ranging from Thar Desert heat to Himalayan winters.

The Print reported that MOSAIC uses an AI and machine learning-based detection technique to distinguish stars from resident space objects (RSOs), including faint targets. Its reported workflow then applies a lost-in-space attitude-estimation approach to determine an object's position without prior information or external cueing.

That architecture differs from tasking a conventional telescope against a known target. The Economic Times described MOSAIC as continuously scanning a portion of the sky rather than requiring an operator to direct the system to specific objects. In space-domain-awareness systems, wide-field sensing and local processing can be important for maintaining observations of objects whose locations are uncertain, although detection performance ultimately depends on factors such as optical sensitivity, weather, node placement, and catalog correlation.

The Economic Times reported that Digantara plans to deploy more than 1,000 MOSAIC nodes globally within 24 months. That target is a reported company ambition rather than an announced deployment schedule for individual sites.

The same report identified celestial navigation as a potential MOSAIC application in GPS-denied environments. The concept uses observed star fields and known celestial references to estimate position or orientation, a technique that can complement satellite-navigation systems where GPS signals are unavailable or disrupted.

India Today framed the launch against India's reliance on domestic satellite assets and commercial imagery during Operation Sindoor, reporting that the event highlighted dependence on external data sources for time-sensitive decisions. The Print similarly reported that Digantara linked the need for India-made surveillance technology to the operation.

For ML and data-engineering teams working on sensing systems, networks of this kind make model accuracy only one part of the operational problem. Comparable distributed surveillance deployments depend on calibrated sensor metadata, reliable time synchronization, image-quality monitoring, object-association pipelines, and uncertainty-aware fusion across nodes. Public reporting does not provide MOSAIC benchmark results, false-positive rates, catalog completeness, or details of its training data, leaving those performance measures as key open technical questions.