Satellite Uses AI to Make Autonomous Decisions in Space
▼ Summary
– A satellite has successfully used onboard AI to autonomously decide when and where to capture scientific images in under 90 seconds without human input.
– The technology, called Dynamic Targeting, was tested by NASA’s JPL on a satellite built by Open Cosmos, using an AI processor from Ubotica.
– The AI analyzed preview images for cloud cover, capturing detailed photos only when skies were clear to save time and resources.
– Unlike traditional satellites that collect all data passively, this system filters useless data in real time, improving efficiency for scientists.
– The technology could be expanded to detect wildfires, volcanic eruptions, and severe storms faster from space, building on prior collaborations between NASA, Ubotica, and Open Cosmos.
A groundbreaking satellite has successfully used artificial intelligence to independently choose when and where to take scientific images, marking a major leap in space technology. The entire decision-making process happens in under 90 seconds without any human intervention, showcasing the potential for smarter, more efficient space missions.
Developed by NASA’s Jet Propulsion Laboratory (JPL), the system, known as Dynamic Targeting, was recently tested aboard a compact satellite designed by UK-based Open Cosmos. The satellite features specialized AI hardware from Dublin’s Ubotica, enabling rapid onboard analysis. During the trial, the satellite scanned 500 kilometers ahead of its orbit, capturing a preliminary image. The AI then assessed cloud cover in real time, if conditions were clear, it took a high-resolution photo; if obscured, it moved on, conserving valuable resources.
Ben Smith from JPL emphasized the advantages of this approach, explaining how selectively capturing images maximizes usable data while minimizing wasted effort. Traditional satellites passively record everything below them, flooding ground teams with raw data that requires extensive sorting. Brian Quinn of Ubotica highlighted the inefficiency of current systems, noting that critical events like wildfires or algal blooms often go unnoticed until days later during post-processing.
This innovation isn’t just about filtering clouds, it lays the foundation for faster disaster response. The same AI could identify wildfires, volcanic activity, or severe storms almost instantly from orbit. The recent test follows earlier collaborations between NASA, Ubotica, and Open Cosmos, including AI-powered cloud detection experiments on the International Space Station and the 2024 launch of HAMMER, a satellite combining hyperspectral imaging with machine learning.
By shifting analysis from Earth to space, this technology promises to revolutionize how we monitor our planet, delivering actionable insights with unprecedented speed. Future missions could leverage similar AI to prioritize high-value targets, transforming satellites from passive observers into intelligent decision-makers.
(Source: The Next Web)
