Machine Learning (ML) has the potential to greatly reduce operational costs and enhance the capabilities of future space systems. Emergent is investigating ML for a variety of applications, from space-based, onboard detection of phenomenologies of interest to fault detection.
Spacecraft anomaly detection and prediction based on telemetry data has been adopted for the operational monitoring of the Attitude Control and Power systems for NASA’s Fermi spacecraft.
We are researching efficient ML algorithms for recognition of phenomena in optical and infrared bands. These algorithms can be rapidly deployed on orbit using our Cirrus product.
We are also working to address challenges with space-based ML such as resource constraints and data sparsity. Extending the capabilities of our Cirrus product for applications with distributed and fractionated ML and investigating methods for on-line and semi supervised space-based training.
We are developing imagery generation capabilities integrated with Ascent and camera simulator to support ML model training and object identification. This enables the augmentation of data sets and creation of data sets that would otherwise be difficult to attain.
