School of Computing

Nov 20
14:00 - 15:00
compint: Agata Rozek
Computational Intelligence Group Seminar
Applications of Machine Learning to Asteroid Shape Modelling -- conclusions from NASA Frontier Development Lab


This summer I was involved in the NASA Frontier Development Lab, an intense 8-week study concentrated on tackling topics important to NASA using machine learning tools. During the programme interdisciplinary teams of early career researchers were looking at issues related to planetary defence, space weather, and space resources. The team I was a part of investigated shape modelling of near-Earth asteroids from radar data. These asteroids are the Earth's closest neighbours in space, most accessible by space flight and with a potential for causing a threat to the planet. Even though they are constantly monitored, detailed characteristics, like shapes and sizes, are available for only a selected few. Physical models are required to successfully plan spacecraft missions and set up impact mitigation strategies. Additional incentive is in learning know how our space environment works and evolves. Reconstructing asteroid shapes and spins from radar data is, like many inverse problems, a computationally intensive task. Shape modelling also requires extensive human oversight to ensure that computational methods find physically feasible results. In this talk I will discuss the results of our work at NASA Frontier Development Lab 2017, exploring the application of machine learning tools to the shape modelling task.



Cornwallis South West,
University of Kent,
United Kingdom


Contact: Marek Grzes

School of Computing, University of Kent, Canterbury, Kent, CT2 7NF

Enquiries: +44 (0)1227 824180 or contact us.

Last Updated: 14/08/2015