Machine Learning-Based and AI Powered Satellite Imagery Processing for Global Air Traffic Surveillance Systems

The unprecedented growth of global air traffic has put immense pressure on the air traffic management systems. In light of that, global air traffic situational awareness and surveillance are indispensable, especially for satellite-based aircraft tracking systems. There has been some crucial development in the field; however, every major player in this arena relies on a single proprietary, non-transparent data feed. This is where this chapter differentiates itself. AIS data has been gaining traction recently for the same purpose and has matured considerably over the past decade; however, satellite-based communication service providers have failed to instrument significant portions of the world’s oceans. This study proposes a multimodal artificial intelligence-powered algorithm to boost the estimates of global air traffic situational awareness using the Global Air Traffic Visualization dataset. Two multimodal artificial intelligence agents categorically detect air traffic streaks in a huge collection of satellite images and notify the geospatial temporal statistical agent whenever both modalities are in concordance. A user can fine-tune the multimodal threshold hyperparameter based on the installed detection rate of datasets to get the best satellite-derived air traffic estimates.