Starry sky with some highlighted constellations

01 October 2017 The “Star Trackers - First Contact” competition has finished. Organized via the online Kelvins platform, the competition attracted several teams worldwide who accepted the challenge and worked on a problem of star identification, as posed and studied by researchers at the Advanced Concepts Team, and produced a number of new approaches.

Star identification is a key technology for spacecraft who must rely on it in order to be able to determine their orientation at all times and can be an issue in situations where, for example, a large number of false stars appear on the camera image or the camera itself has unidentified aberrations / alterations.

The aim of the competition was to improve on the current state of the art by asking participants to produce algorithms that exhibit robustness to false stars and unforeseen camera aberrations. Algorithms able to perform the identification correctly and in a short time were ranked higher.

An international team including researchers from the Instituto Nacional de Pesquisas Espaciais (INPE, São José dos Campos, Brazil), the Centro Universitario de la Defensa - Zaragoza (CUD, Spain), and Texas A&M University (TAMU, College Station, USA), submitted a new algorithm called Super k-ID which resulted to be both the most accurate and fast, de facto winning the competition.

A second algorithm revealed to have very competitive performances. The algorithm is called “Multi Poles Algorithm” and was developed by the Automation, Robotics and Control for Aerospace lab (ARCAlab) of the School of Aerospace Engineering, “Sapienza” University of Rome, Italy. The Multi Poles Algorithm (MPA) was modified and improved for the contest. Other algorithms also produced remarkable results and advanced significantly the status quo. The complete summary of the final results is also available here

Researchers at the Advanced Concepts Team had worked on star-trackers algorithms before releasing the competition studying a small but important part of robust approaches to star identification during the study Optimal orderings of k-subsets for star identification during which the idea for the competition originated.

Contacts:

Advanced Concepts Team: Dario Izzo