Abstract:Since the launch of Sputnik in 1958, thousands of satellites have been launched by governments and commercial entities for a wide range of purposes. Once a satellite has completed its primary mission, most operators de-orbit or move the satellite into a graveyard orbit as to avoid collisions with active satellites. However, some spacecraft do not have any residual propellant. Other satellites suffer failures and become hazards themselves. Non-operational satellites in orbit have become an increasing collision risk for operational satellites and this problem will only become worse as the number of satellites launched every year increases. Although finding ways to clean up orbital debris is important, near term solutions is needed to mitigate the risk of debris colliding with an active satellite. Traditionally, debris avoidance is performed by operators on the ground who calculate debris conjunctions and analyze the risk of moving the satellite. This paper proposes an alternate solution where the process of analyzing and avoiding orbital debris can be done autonomously by the satellite without human intervention. This provides advantages such as reducing the number of operators needed to maintain a satellite or a large constipation of satellites thus reducing operating costs. With the increased amount of processing power on modern satellites, this system is possible with current technology. The system would first collect a database of two line element's (TLE) for each piece of orbital debris. The satellite will then perform a conjunction analysis against each piece of debris and the satellites predicted future position. The satellite will re-assess these probabilities at set increments as it gets closer to the conjunction, or when the debris database is updated. The satellite operator will be able to set the collision probability threshold. If this threshold is exceeded, the satellite will calculate the most efficient maneuver based off the required delta v, time to the conjunction and staying in its operational orbit. The system will also attempt to combine any required avoidance burns with a normal orbit maintenance burn to minimize fuel consumption. When an appropriate avoidance burn is being considered, the system will check for debris conjunctions in this new orbit. If there is no danger of a collision, the maneuver will be executed.

Learned Prediction for Space Debris Collision Risk and Avoidance

Predicting the Probability of Collision of a Satellite with Space Debris: A Bayesian Machine Learning Approach

Reentry Risk and Safety Assessment of Spacecraft Debris Based on Machine Learning

A Population Dynamics Approach to the Distribution of Space Debris in Low Earth Orbit

An Efficient Algorithm for Collision Avoidance Between a Solar Array Satellite and Space Debris

SPACE DEBRIS COLLISION PREDICTION RESEARCH

Machine learning for the prediction of local asteroid damages

Probability of Collision of satellites and space debris for short-term encounters: Rederivation and fast-to-compute upper and lower bounds

Reduction of Space Debris Collision Prediction Uncertainty Based on Q-Sat Precise Orbit

Spacecraft Collision Avoidance Challenge: design and results of a machine learning competition

Recurrent Convolution Neural Network for Space Debris and Asteroids Detection and Classification on the Monitored Satellite Information Using Deep Space Network

Machine Learning in Orbit Estimation: a Survey

An investigation on space debris of unknown origin using proper elements and neural networks

A PCA-SVM Regression Model for LEO Space Debris Orbit Prediction in an Optical Space-Based Space Surveillance Network

An autonomous satellite debris avoidance system

Collision Risk Assessment for a Spacecraft in the Space Debris Environment

Machine learning-based classification for Single Photon Space Debris Light Curves

Density-based in-orbit collision risk model valid for any impact geometry

Towards Automated Satellite Conjunction Management with Bayesian Deep Learning

Predicting the Position Uncertainty at the Time of Closest Approach with Diffusion Models

A deep neural network framework with Analytic Continuation for predicting hypervelocity fragment flyout from satellite explosions