AI-Powered Insights
LeoLabs applies artificial intelligence (AI) and machine learning (ML) throughout our technology stack, from optimizing our Global Radar Network to detecting events on orbit and identifying abnormal spacecraft behavior based on patterns of life.
AI/ML for Space Domain Awareness
Our advanced analytic tools for orbital activity can seamlessly integrate with existing systems for defense and intelligence customers. We employ an automated retraining pipeline to ensure our AI/ML models keep pace with new launches and changes to routine object behavior.
Maneuver Detection & Characterization
Detects maneuvers accurately and in real-time by estimating natural decay due to atmospheric drag and changes in velocity and altitude, then flags orbital state changes that exceed those thresholds. This model also characterizes the time, number, and types of maneuvers among all spacecraft in LEO.
Patterns of Life
Analyzes all objects in our expansive catalog daily to summarize orbital trends and highlight objects exhibiting unusual activity to assign an “interest score.” Highest-scoring objects are flagged for further analysis and reporting.
Multi-Object Detection
Detects new objects that result from deployments, collisions, or explosions with a low rate of false positives.
Object Characterization
This multi-model system characterizes objects with unknown properties and derives insights from object relationships. Outputs include object and payload maneuverability classification, hard body radius estimates, and visualization of object similarities across the LeoLabs catalog.
AI/ML for Radar Scheduling
As the number of objects in LEO rapidly increases, LeoLabs radars commonly have multiple objects in their field of view at a given time. Our dynamic, intelligent scheduling system tells the radars which objects to prioritize and how long to track a specific object for. This technology enables us to track objects 40% more frequently and with greater accuracy.
Probability of Detection
This supervised model optimizes the relative priority of space objects for radar scheduling and tracking by forecasting the probability of collecting a measurement on that object for a given radar pass and the operational utility that would result from it. If multiple objects are in a radar’s field of view, higher priority objects are tracked first.
Beam Length
This model, also supervised, dynamically adjusts the length of time an object is tracked during a given radar pass. This increases efficiency by reducing the average amount of radar time needed to collect measurements on each passing object.