Probabilistic Space Debris Conjunction Assessment using Machine Learning and Covariance Intersection Analysis
Year: 2026
Type: conference-paper
Abstract: The exponential proliferation of artificial satellites and the accumulation of micro-debris in Low Earth Orbit presents a critical threat to modern space infrastructure, commonly referred to as the Kessler Syndrome. Standard deterministic physics models frequently fail to predict orbital conjunctions accurately due to inherent uncertainties in radar telemetry and state vectors. This paper presents a novel, real-time tracking and prediction system that bridges high-fidelity orbital mechanics with machine learning. Utilizing a dual-engine WebGL architecture, the system propagates dynamic Earth-Centered Inertial coordinates to Geodetic spatial mappings at high frame rates using live Two-Line Element data. To address the limitations of deterministic distance calculations, a machine learning pipeline is integrated to evaluate the probability of collision and covariance intersection matrices. The accuracy of the physics engine and the necessity of the predictive model are validated through a time-decoupled historical reconstruction of the 2009 collision between the Iridium 33 and Cosmos 2251 satellites. Furthermore, the system introduces a GPU-accelerated debris swarm simulation and volumetric density heatmaps to visualize orbital congestion zones. By exporting time-series kinematic data for analysis, this system provides a scalable, enterprise-grade framework for both real-time operational tracking and academic research in space situational awareness.
Institution: Silver Oak University
Language: English
Subfield: Aerospace Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7204141591