SPACE OBJECT CLASSIFICATION USING MODEL DRIVEN AND DATA DRIVEN METHODS

In recent years there has been an increase in the number of inactive and debris Space Objects (SOs). This work examines both data driven and model driven SO classification. The model driven approach investigated for this work is based on the Multiple Model Adaptive Estimation approach to extract SO characteristics from observations while estimating the probability the observations belonging to a given class of objects. The data driven methods are based on Principle Component Analysis and Convolutional Neural Network Classification approaches. The performance of these strategies for SO classification is demonstrated via simulated scenarios.