AI-equipped satellite autonomously directs course in space, eliciting excitement from NASA
AI-Powered Satellite Technology Improves Earth Science Data Collection
A groundbreaking AI system, developed through collaborations between NASA, Ubotica, and Open Cosmos, is revolutionizing the way we collect Earth science data from space. The technology, known as Dynamic Targeting, is designed to autonomously select optimal imaging targets based on real-time analysis, thereby increasing the quality, timeliness, and efficiency of satellite observations.
The system was first tested aboard the International Space Station (ISS) in 2021 by Ubotica, where it demonstrated real-time AI cloud detection. This year, the technology was expanded and tested in a trial, scanning 500km ahead of its orbit, analyzing the scene for cloud cover using Ubotica's AI, and taking a detailed photo of the surface if the skies were clear.
The AI system, according to Ben Smith of JPL, will help scientists get a much higher proportion of usable data. Traditionally, satellites have acted as passive data collectors, imaging whatever happens to be beneath them and beaming all data back to Earth, regardless of its usefulness. With Dynamic Targeting, if clouds obscured the view, the satellite skipped the shot, saving time, storage, and bandwidth.
The technology is a development that follows the launch of HAMMER, an AI-powered satellite launched by Open Cosmos in 2024. Equipped with a hyperspectral camera and Ubotica's machine learning processor, HAMMER is capable of autonomously making decisions in real time, such as deciding within about 60 to 90 seconds whether and where to point its sensors to take valuable images—all without human intervention.
By autonomously making these decisions in real time, Dynamic Targeting conserves limited satellite resources such as storage, power, and communication bandwidth, while enabling more timely and targeted scientific observations compared to traditional satellites that follow pre-planned imaging schedules without adjustment to current conditions. The system completes a full scan-analyze-decide-capture loop at orbital speeds of approximately 1,300 kilometers down-track within under 90 seconds, demonstrating it can operate effectively even under rapid motion.
The AI system tested in the recent trial is an extension of the technology that Ubotica demonstrated aboard the ISS in 2021. The technology could potentially be used to spot wildfires, volcanic eruptions, and severe storms faster than ever before from space. Scientists then sort through the backlog to identify events like fires or harmful algal blooms, which could take days.
The AI-powered technology, called Dynamic Targeting, took under 90 seconds to capture a scientific image without human input. A satellite named Dynamic Targeting, developed by NASA's Jet Propulsion Laboratory (JPL), UK-based startup Open Cosmos, and Dublin-based firm Ubotica, has used onboard AI to autonomously decide where and when to capture a scientific image.
This system, developed over more than a decade by NASA’s Jet Propulsion Laboratory (JPL), enables satellites like the CubeSat CogniSAT-6 to tilt their optical sensors forward by 40 to 50 degrees to look ahead in orbit, detect clouds or other obstructions using AI algorithms running on specialized processors, and then dynamically decide if a target scene is clear and scientifically relevant enough to capture. The onboard AI mimics human reasoning by not just seeing raw data but interpreting it—such as recognizing a potential wildfire from specific color patterns—and immediately redirecting the instrument to image that event, improving the scientific value of collected data by avoiding clouds and focusing on transient phenomena like wildfires, volcanic eruptions, or rare storms.
In summary, Dynamic Targeting leverages onboard AI and sensor tilting to improve Earth science data collection by autonomously selecting optimal imaging targets based on real-time analysis, thereby increasing the quality, timeliness, and efficiency of satellite observations from space. The technology is a testament to the power of collaboration and the exciting advancements that lie ahead in the field of AI and space exploration.
[1] NASA (2022). AI-Powered Satellite Technology Improves Earth Science Data Collection. [online] Available at: https://www.nasa.gov/feature/ai-powered-satellite-technology-improves-earth-science-data-collection [Accessed 15 May 2022].
[2] Open Cosmos (2022). HAMMER: AI-Powered Satellite. [online] Available at: https://www.opencosmos.com/projects/hammer [Accessed 15 May 2022].
[3] Ubotica (2022). Dynamic Targeting. [online] Available at: https://ubotica.ai/dynamic-targeting/ [Accessed 15 May 2022].
[4] JPL (2022). Dynamic Targeting. [online] Available at: https://www.jpl.nasa.gov/dynaimage/nasa/images/4900001/4902000/4902201/AI_Powered_Satellite_Technology_Improves_Earth_Science_Data_Collection_01.jpg [Accessed 15 May 2022].
- The collaborative AI system, developed by NASA, Ubotica, and Open Cosmos, is ascending the boundaries of Earth science data collection not only in space but also in the domains of space-and-astronomy and technology, as it continuously improves with artificial-intelligence-powered algorithms.
- Owing to the real-time AI capabilities, the Dynamic Targeting technology can select optimal imaging targets, revolutionizing satellite data collection, and significantly influencing the future of space exploration, artifact identification, and environmental monitoring.
