Satellite density in Earth orbit has increased exponentially in recent years, with the lower cost of small satellites allowing governments, researchers and private companies to launch and operate an estimated 2,877 satellites in orbit by 2023 alone. This includes increased activity of satellites in geostationary orbit (GEO), which bring technologies with global impact, from broadband internet to climate monitoring. However, along with the multiple benefits of these satellite-based technologies come increased safety and security risks, as well as environmental concerns. More accurate and efficient methods of monitoring and modeling satellite behavior are urgently needed to prevent collisions and other disasters.
To address this challenge, MIT's Astrodynamics, Space Robotics, and Controls Laboratory (ARCLab) launched the ai/web/challenges/challenge-page/2164/overview”>MIT ARCLab Award for Innovation in artificial intelligence in Space: a first-of-its-kind competition that asks participants to use ai to characterize satellite patterns of life (PoL)—the long-term behavioral narrative of a satellite in orbit—using purely passively collected data. Following a call for entries last fall, 126 teams used machine learning to create algorithms to label and timestamp GEO satellites’ behavioral modes over a six-month period, competing for accuracy and efficiency.
Supported by the U.S. Department of the Air Force and the MIT ai Accelerator, the challenge offers a total of $25,000. A team of judges from ARCLab and MIT Lincoln Laboratory evaluated the proposals based on clarity, novelty, technical depth, and reproducibility, assigning each proposal a score out of 100 points. Now, the judges have announced the winners and runners-up:
First prize: David Baldsiefen — Team Hawaii2024
With a winning score of 96, Baldsiefen will receive a $10,000 prize and be invited to join the ARCLab team to present a poster session at the Maui Advanced Optical and Space Surveillance Technologies (AMOS) Conference in Hawaii this fall. One reviewer noted, “Clear and concise paper, with very good ideas such as localizer tag encoding. Decisions on architectures and feature engineering are well reasoned. The code provided is also well documented and structured, allowing for easy reproducibility of experimentation.”
Second Prize: Binh Tran, Christopher Yeung, Kurtis Johnson, Nathan Metzger — Millennial-IUP Team
With a score of 94.2, Y, Millennial-IUP will receive $5,000 and join the ARCLab team at the AMOS conference. One reviewer said: “The models chosen were sensible and justified, making impressive efforts in efficiency gains… They used physics to inform their models and this appeared to be reproducible. Overall, it was an easy to follow, concise, and jargon-free paper.”
Third prize: Isaac Haik and Francois Porcher — QR_Is Team
With a score of 94, Haik and Porcher will share the third prize of $3,000 and will also be invited to the AMOS conference with the ARCLab team. One reviewer noted: “This informative and interesting paper describes the combination of ML and signal processing techniques in a compelling way, aided by informative graphs, tables, and sequence diagrams. The author identifies and describes a modular approach to class detection and its assessment of feature utility, which they correctly identify as not being uniformly useful across classes… Any lack of mission experience is made up for by a clear and detailed discussion of the benefits and drawbacks of the methods they used and a discussion of what they learned.”
Teams placing fourth through seventh will each receive $1,000 and a certificate of excellence.
“The goal of this competition was to foster an interdisciplinary approach to problem solving in space by inviting experts in ai development to apply their skills in this new context of orbital capability. And all of our winning teams truly delivered on their promise – bringing technical skills, novel approaches and experience to a very impressive round of submissions,” says Professor Richard Linares, who leads ARCLab.
Active modeling with passive data
During the time a GEO satellite remains in orbit, operators issue commands to place it into various behavior modes: station keeping, longitudinal shifts, end-of-life behaviors, etc. Satellite Patterns of Life (PoL) describe in-orbit behavior composed of sequences of both natural and unnatural behavior modes.
ARCLab has developed an innovative benchmarking tool for characterizing patterns of life on geostationary satellites and created the Satellite Pattern of Life Identification Dataset (SPLID), which comprises data from real and synthetic space objects. Challenge participants used this tool to create algorithms that use ai to map a satellite’s in-orbit behaviors.
The goal of the MIT ARCLab Prize for Innovation in ai in Space is to encourage technologists and enthusiasts to bring innovation and new skill sets to established challenges in the aerospace sector. The team intends to hold the competition in 2025 and 2026 to explore other topics and invite ai experts to apply their skills to new challenges.
