COMPUTATIONAL INTELLIGENCE AND
THE QUANTIFICATION OF UNCERTAINTY
State of the Art
The prediction of rare and catastrophic events, like collisions with space debris or impacts with asteroids, and anomalies that can lead to catastrophic events is key to improve the resilience of the space environment and planetary defence. This requires an element of uncertainty quantification and an element of operation planning and optimization. In both contexts, computational intelligence can offer machine learning, optimisation and statistical analysis techniques to study the global behaviour of space objects, identify rare events and quantify the associate uncertainty. It can also be used to provide optimal planning and scheduling of collision and impact avoidance manoeuvres, fundamental in Space Traffic Management and Planetary Defence. The problem of correlating spatially and temporally distant events translates into processing multiple sources of information to learn from past data about patterns and early warning signals that can predict rare events and plan recovery measures. Data fusion techniques able to deal with imprecise, incomplete and spurious data have been lately subject of an intense research and techniques coming from the field of fuzzy set theories, probabilistic methods and evidence theory have been applied also in this context. Moreover recent advances in deep belief networks for unsupervised anomaly detection, and time series analysis for anomaly prediction, have proved to be more effective in modelling of complex nonlinear systems.
WP1 combines mathematical modelling with artificial and computational intelligence to quantify uncertainty in orbital mechanics, predict and correlate rare events, anomalies and singularities, and support decision making and operation planning. It covers the computation of the probability of impacts and collisions, with single objects and clouds of fragments, including resonant returns and keyholes, and the quantification of uncertainty during re-entry.
Manzi, M., & Vasile, M. (2020). Discovering unmodeled components in astrodynamics with symbolic regression. 2020 IEEE Congress on Evolutionary Computation (CEC), Glasgow, United Kingdom, 19-24 July 2020, pp. 1-7.
Manzi, M., & Vasile, M. (2020). Analysis of stochastic nearly-integrable dynamical systems using polynomial chaos expansions. 2020 AAS/AIAA Astrodynamics Specialist Conference, South Lake Tahoe, United States, 09-12 August 2020.
Manzi, M., & Vasile, M. (2020). Asteroid Deflection Under Uncertainty. Stardust-R Global Virtual Workshop I, Pisa, Italy, 07-10 September 2020.
Manzi, M., & Vasile, M. (2020). Orbital anomaly reconstruction using deep symbolic regression. 71st International Astronautical Congress, IAC, Dubai, United Arab Emirates, 12-14 October 2020.
Stevenson, E., Rodriguez-Fernandez, V., Minisci, E., Camacho, D. (2020). A Deep Learning Approach to Space Weather Proxy Forecasting for Orbital Prediction. In Proceedings of the 71st International Astronautical Congress (IAC), The CyberSpace Edition, 12-14 October 2020.
Stevenson, E., Rodriguez-Fernandez, V., Urrutxua, H., Morand, V., Camacho, D. (2021). Artificial Intelligence for All vs. All Conjunction Screening. In Proceedings of the 8th European Conference on Space Debris, 20-23 April 2021. (Accepted).