Satellite Achieves Landmark: Onboard AI Autonomously Directs Scientific Imaging
A new era in space-based observation has begun, with a satellite successfully using artificial intelligence to independently determine when and where to capture scientific images – all within 90 seconds and without any human intervention.
For the first time, a satellite has demonstrated the ability to make real-time decisions about data acquisition, a capability poised to revolutionize Earth observation and disaster response. The breakthrough, achieved earlier this month, was the result of a collaborative effort between NASA’s Jet Propulsion Laboratory (JPL), UK-based startup Open Cosmos, and Dublin-based firm Ubotica.
Dynamic Targeting: A New Approach to Space-Based Data Collection
The technology, dubbed Dynamic Targeting, was tested on a satellite built and operated by Open Cosmos. The satellite utilizes a machine learning processor developed by Ubotica, enabling it to analyze potential imaging targets before committing to a capture. In the recent trial, the satellite scanned 500 kilometers ahead of its orbit, taking a preliminary image.
Ubotica’s AI swiftly analyzed the scene for cloud cover. If conditions were clear, the satellite reoriented itself to acquire a high-resolution image of the surface. Conversely, if clouds were present, the satellite bypassed the image capture, conserving valuable resources like time, storage capacity, and bandwidth.
“If you can be smart about what you’re taking pictures of, then you only image the ground and skip the clouds,” a senior official at JPL stated. “This technology will help scientists get a much higher proportion of usable data.”
From Passive Collectors to Intelligent Observers
Traditionally, satellites have functioned as largely passive data collectors, indiscriminately imaging everything within their field of view and transmitting all data back to Earth for subsequent analysis. This process often results in a significant backlog of unusable data, requiring extensive post-processing by scientists.
“It takes post-processing, which could be days later, to say, ‘Hey, there was a fire. Hey, there was a harmful algal bloom’,” explained a company representative from Ubotica in a recent NASA publication. Dynamic Targeting flips this paradigm, enabling satellites to act as intelligent observers, prioritizing valuable data and reducing wasted resources.
Expanding Capabilities: Beyond Cloud Detection
The potential applications of this technology extend far beyond simply avoiding cloudy images. NASA, Ubotica, and OpenCosmos envision the system being adapted to rapidly identify and monitor wildfires, volcanic eruptions, and severe storms from space. This could dramatically improve early warning systems and response times for natural disasters.
This latest success builds upon a history of collaboration between the three organizations. In 2021, Ubotica demonstrated real-time AI cloud detection capabilities aboard the International Space Station (ISS), as part of a broader research initiative with JPL. Further advancing the technology, Open Cosmos launched HAMMER in 2024 – an AI-powered satellite equipped with a hyperspectral camera and Ubotica’s machine learning processor.
The development of Dynamic Targeting represents a significant leap forward in space-based technology, paving the way for a future where satellites are not just collecting data, but actively contributing to our understanding of a rapidly changing world.
