Idea
A predictive modeling platform for satellite operators and space agencies to forecast LEO object populations efficiently and accurately.
Research Paper
Core Innovation
This paper introduces a hybrid modeling approach combining Sparse Identification of Nonlinear Dynamics (SINDy) and Long Short-Term Memory (LSTM) networks to replicate high-fidelity LEO orbit population dynamics. It uniquely balances accuracy and computational efficiency by training on data from the MOCAT-MC model, enabling faster predictions without sacrificing precision. This approach advances orbit capacity modeling beyond traditional computationally expensive simulations.
Market Size (TAM)
$2–10B TAM, $1–2B SAM; assumption: growing satellite deployment and increasing demand for space traffic management tools.
Potential Customers & Pain Points
- Satellite Operators Needing Efficient Orbit Capacity Forecasts
- Space Agencies Managing Space Traffic and Debris
- Aerospace Companies Reducing Computational Costs for Orbit Modeling
Business Model
Subscription-based SaaS platform offering API access to predictive orbit capacity models and custom forecasting services.
Competitive Landscape
- AGI Systems Tool Kit
- LeoLabs
- ExoAnalytic Solutions
Implementation Challenges
- Access to High-Quality Training Data
- Integration with Existing Space Traffic Systems
- Regulatory and Compliance Challenges
Validation Strategy
- Benchmark model predictions against MOCAT-MC outputs
- Pilot deployment with satellite operators for real-world testing
- Iterate model based on user feedback and accuracy metrics
Research Paper Overview
A Data-Driven Approach to Estimate LEO Orbit Capacity Models
Summary
This paper uses SINDy and LSTM algorithms to model and predict the population dynamics of resident space objects in LEO, including Active, Derelict, and Debris categories. It leverages data from the high-fidelity MOCAT-MC model to create a faster, low-fidelity forecasting tool that maintains accuracy while reducing computational cost.