RESIDENT SPACE OBJECTS DETECTION AND TRACKING BASED ON ARTIFICIAL INTELLIGENCE

Resident Space Objects (RSOs) detection and tracking are challenging problems in the framework of Space Situational Awareness (SSA). The growing number of in orbit platforms and the incoming era of mega constellations is increasing both active and passive traffic in the near Earth segment. Recently, more and more researchers and companies have started investigating the problem. This, combined with the growing popularity of Artificial Intelligence (AI) applications, has led to interesting solutions. The present work will investigate an AI based approach for Image Processing, Objects Detection and Tracking oriented towards space optical sensors applications. It will show the architecture development and test of a Convolutional Neural Network (CNN) based algorithm: the image processing and object detection tasks are demanded to Neural Network (NN) modules while the tracking of objects inside the sensor’s Field Of View (FOV) is formulated as an optimization problem. Dataset creation for the network training, algorithm design process and results both on real and simulated images will be shown.