Predictive & Risk-Aware

Space Domain Awareness

View Full Animation →
Startup Philosophy

Simple → Interpretable → Safe

We design safety-critical systems so they remain simple enough to audit, interpretable enough to trust, and safe enough to deploy.

Simple

Constraint-first models, bounded complexity, and physics-guided structure keep the system understandable from the start.

bounded models · physics-guided · constrained

→

Interpretable

Clear uncertainty estimates, traceable model behavior, and auditable outputs make decisions explainable.

uncertainty bounds · traceability · auditable

→

Safe

Operational guardrails, calibrated risk signals, and verification-oriented design support safer deployment in critical environments.

guardrails · calibrated risk · verification

Safety emerges when system behavior stays inspectable under operational pressure.

THE CHALLENGE:
CATALOG SCALING

Tens of thousands of space objects require multi-day conjunction screenings. High-fidelity uncertainty propagation is computationally prohibitive, causing false alerts and operator overload.

Challenge visualization

THE SOLUTION:
CTPC AUGMENTATION

A lightweight, Machine Learning-augmented probabilistic correction layer that snaps into existing physics-based propagators to provide statistically calibrated uncertainty.

Solution visualization

THE IMPACT:
OPERATIONAL SAFETY

Reduces false alerts, mitigates probability dilution, and enables automated risk-aware maneuver planning without disrupting current operational pipelines.

Impact visualization
SDA PLATFORM ARCHITECTURE

From Observations to Maneuver Decisions

An end-to-end architecture where thermospheric density drives drag uncertainty, CTPC calibrates covariance in real time, and operators receive stable, actionable collision risk — not volatile alerts.

1 Observation Layer
Sensors
→
Tracking Networks
→
Orbit Determination
↓
2 Environment Layer SiS
Space Weather Inputs F10.7 • Kp/Ap • Solar wind/IMF
→
Density Modeling NRLMSISE-00 / JB2008 • FNO / SH
→
Density Field Output Spatio-temporal thermospheric density • Orbit-sampled profiles
Feeds into propagation drag forcing + CTPC environmental conditioning
↓ drag forcing ↓ env signals
3 Prediction Layer
Deterministic Propagation GMAT / physics model + env-aware drag
→
CORE CTPC Probabilistic Corrector Environment-Conditioned Uncertainty Calibration
trajectory history covariance history environmental signals
↓
4 Uncertainty Layer SiS
Calibrated Covariance Σ̃
→
Reliability Metrics Mahalanobis consistency • Coverage • Sharpness
Uncertainty is conditioned on space weather regime
↓
5 Risk Intelligence Layer
Collision Probability Pc
→
Pc Stability Analysis
→
Alert Volatility Reduction

Stable risk estimates enabled by calibrated uncertainty

↓
6 Decision Layer
NEW Risk-Aware Maneuver Optimization CVaR / probabilistic constraints • fuel vs safety
→
Decision Consistency Monitoring
→
Operator Decision Support
↺ Closed-loop decision-aware prediction — feeds back to Propagation + Environment interaction ↻
Environment-aware uncertainty calibration stabilizes collision risk estimation
and enables reliable maneuver decisions.
Existing SDA Component
SiS Core Innovation (CTPC + Environment)
Decision Output

This is an environment-conditioned probabilistic system that transforms uncertainty into a decision-grade signal.

Work With Us

We are building safety-critical AI systems for space operations. If you are working on orbital safety, space traffic management, or mission-critical autonomy, we would love to collaborate.

Contact Us Research Collaboration

Based in Ames, Iowa. Working at the intersection of machine learning, orbital dynamics, and safety-critical systems.