Enhanced Trajectory Prediction of Satellites and Space Debris Using Machine Learning and Kalman Filter

The increasing proliferation of space debris poses a significant and growing threat to operational spacecraft and future space missions. Our research aims to address the growing concern about space debris and the need for accurate trajectory predictions to ensure the safety and sustainability of space operations. Our approach combines machine learning for space object classification with classical filtering techniques for trajectory prediction, resulting in an interactive visualization of the spatial environment. The initial phase of our research consisted of applying a Random Forest Classifier for the accurate detection and classification of space objects, distinguishing between active satellites and space debris. Subsequently, our research used a Kalman filter to predict the trajectories of both active satellites and space debris. This allowed us to obtain dynamic and precise position informations for these space objects. Finally, a 3D visualization has been developed to illustrate the behavior and movement of both debris and active satellites. Preliminary results, obtained by extracting orbital parameters such as semi-major axis, inclination, right ascension of the ascending node (RAAN), argument of perigee, and mean anomaly from Two-line element (TLE) data, indicated a good classification accuracy of approximatively 98% for distinguishing between different types of space objects during the training phase.