The number of satellites orbiting the Earth has grown exponentially in recent years, both due to lower costs and a rise in demand for services that satellites can provide, such as broadband internet and climate surveillance. However, this increase in activity also raises concerns around safety, security, and the environment, necessitating enhanced methods for monitoring and modeling satellite behavior to prevent accidents and collisions.
To tackle this issue, the MIT Astrodynamics, Space Robotic, and Controls Laboratory (ARCLab) initiated the MIT ARCLab Prize for AI Innovation in Space. This pioneering competition invites participants to use artificial intelligence (AI) to outline and characterize satellite behavior patterns, using only passively gathered information.
Following the launch of the competition, 126 teams used machine learning algorithms to determine and assign timestamps to the behavioral patterns of geostationary Earth orbit (GEO) satellites over six months. These selections were judged on clarity, originality, technical depth, and reproducibility, with judges from ARCLab and MIT Lincoln Laboratory grading each entry on a 100-point scale.
The winning entry, Team Hawaii2024, scored 96 points and won $10,000. Team Millennial-IUP came in second with a score of 94.2 and a prize of $5,000. Team QR_Is finished third, with a score of 94 and earning a prize of $3,000. The fourth- to seventh-place teams each received $1,000 and a certificate of excellence.
The competition was designed to stimulate an interdisciplinary problem-solving approach within the realm of space, by tempting AI experts to apply their talents. The challenge involved mapping out various behavioural modes of GEO satellites, which operators command based on both natural and non-natural factors.
ARCLab has created an innovative benchmarking tool for determining geosynchronous satellite behavioural patterns, called the Satellite Pattern-of-Life Identification Dataset (SPLID). This database comprises both real and synthetic space object data, and it was used by the contest participants to build out their algorithms.
The goal of the MIT ARCLab Prize for AI Innovation in Space is to encourage AI technological enthusiasts to contribute innovative solutions to established issues in aerospace. The team plans to continue the competition in 2025 and 2026, which will focus on other issues while encouraging AI experts to further use their talents in entering new challenges.
