Over a dozen years of AI in Space on Earth Observing One and Sensorwebs
Presenting Author: Steve Chien
Organization: Jet Propulsion Laboratory, Cal Inst Tech / NASA
Co-Author(s): S. Chien, G. Rabideau, D. Tran, S. Schaffer, A. Davies, R. Greeley, V. Baker, S. Frye, D. Mandl
AI software operated the Earth Observing One Mission for over a dozen years. Numerous technologies were demonstrated as part of the Autonomous Sciencecraft and used operationally on Earth Observing One. First, many data interpretation algorithms including Decision Trees, Support Vector Machine Learning, Random Decision Forests, Bayesian Thresholding, Unsupervised Learning, and even Hyperspectral Unmixing. Second, automated scheduling technologies were used operationally both ground and onboard to enable autonomous operations. Finally, EO-1 was networked with multiple other space, ground, air, and marine assets in an ad hoc sensorweb to track numerous phenomena including volcanoes, flooding, wildfires, cryosphere and more. We discuss technologies used, lessons learned, and future directions of work.