AI-based data fusion for orbital object data

Public space object data released by the US Space Force is a fantastic resource. Sadly, it is not accurate enough to provide any kind of operational space safety service. 

Through our paper “Systematic TLE data improvement by neural network for most catalogued resident space objects” authored by Hilaire Bizalion (Ecole Polytechnique), you will understand why it is, indeed, a bad idea. But don’t worry, we also explain how you can use neural networks to make the data much more friendly in terms of statistics, and also shrink the error bars by a factor 5. 

Thank you to the COSPAR for featuring Share My Space in Advances in Space Research, one of the leading journals in space sciences. 

https://www.sciencedirect.com/science/article/abs/pii/S0273117722005804

‍