Project Overview
This research project focuses on developing advanced orbit determination techniques using the novel Higher-Order Unscented Kalman Estimator (HOUSE) framework. The work addresses critical challenges in space surveillance and tracking, particularly for satellites in low Earth orbit (LEO).
Key Achievements
Square-Root HOUSE (w-HOUSE) Development
Our breakthrough research has led to the development of the w-HOUSE filter, which represents a significant advancement in nonlinear filtering for orbit determination applications. This work was published in IEEE Transactions on Aerospace and Electronic Systems in 2024.
Performance Improvements
The w-HOUSE filter demonstrates:
- Superior Accuracy: 3D root-mean-square errors of less than 60 metres over three-day tracking scenarios
- Enhanced Robustness: Better handling of outlier-contaminated measurements compared to traditional methods
- Computational Efficiency: Square-root formulation ensures numerical stability
Technical Approach
Challenges Addressed
- Limited Measurement Time: Short visible arcs from ground-based optical tracking
- Non-Gaussian Data: Presence of outliers in observational data
- Nonlinear Dynamics: Highly perturbative orbit dynamics in LEO environment
- Sparse Observations: Irregular measurement availability
Methodology
- Higher-Order Unscented Transformation: Enhanced sigma point selection for better nonlinear approximation
- Square-Root Implementation: Improved numerical stability and computational efficiency
- Robust Parameter Estimation: Advanced techniques for handling measurement uncertainties
- Multi-Sensor Integration: Fusion of angle-only measurements with other sensor data
Validation and Results
Real-World Testing
The algorithm has been validated using:
- Synthetic Datasets: Comprehensive Monte Carlo simulations
- Real Measurements: Optical tracking data from Sentinel-6A satellite
- Comparative Analysis: Performance benchmarking against UKF, CUT filters, and precise GNSS-based solutions
Performance Metrics
- Positioning accuracy: < 60m RMS error over 3 days
- Robustness to outliers: 30% improvement over traditional methods
- Computational efficiency: Real-time capable implementation
Applications
This research has direct applications in:
- Space Situational Awareness: Enhanced tracking of space debris and active satellites
- Mission Operations: Improved spacecraft navigation and collision avoidance
- Scientific Missions: Precise orbit determination for Earth observation satellites
- Commercial Space: Supporting the growing commercial satellite industry
Code Availability
The implementation of the HOUSE algorithms is available on GitHub, supporting reproducible research and enabling the space community to benefit from these advances.
Future Work
Ongoing developments include:
- Extension to higher-altitude orbits
- Multi-object tracking capabilities
- Integration with machine learning approaches
- Real-time implementation on space-qualified hardware