The accurate prediction of the re-entry of objects orbiting around the Earth is crucial for the safety of people and properties on the ground. In our latest work, we explore the possibility of paradigm shift: from a dynamics-based approach to a data-driven approach for re-entry prediction. Feeding a deep learning Recurrent Neural Network architecture with TLE data, we can provide accurate prediction of the re-entry time of uncontrolled objects.
The paper is available with free access at http://bit.ly/3Tn7UDS
