NASA is testing a cutting-edge form of onboard artificial intelligence that could significantly transform how satellites monitor Earth. In a recent demonstration, a satellite independently detected clouds ahead, processed that information itself, and, in under 90 seconds, chose whether to photograph the ground or skip the shot, no directions from ground control required.

Dynamic Targeting: Smarter Satellite Decisions

This AI system, called Dynamic Targeting, was developed at NASA’s Jet Propulsion Laboratory (JPL) over the past decade. It represents a big step toward making satellites more autonomous. As Steve Chien, JPL’s technical fellow in AI and the project’s lead, explains, the aim is for satellites to analyze what they see and decide how to react, much like a human would. The initial focus is helping satellites distinguish clear skies from clouds so they can avoid wasting bandwidth and storage on unusable, cloud-obstructed images.

How It Works: Cloud Avoidance in Real Time

The first real-world test took place aboard CogniSAT-6, a compact CubeSat launched in March 2024. Built by Open Cosmos and using an AI processor from Ubotica, this spacecraft showed Dynamic Targeting in action: spotting clouds and changing its behavior accordingly. Without a dedicated camera for looking ahead, the CubeSat tilts itself 40–50° to peek ahead along its orbit using its optical sensor, which sees both visible and near-infrared light. The onboard AI runs an algorithm trained to spot clouds; if skies are clear, the satellite gets ready to image the ground, but if clouds are present, it cancels the shot, saving power and storage.

According to Ben Smith from JPL’s Earth Science Technology Office, this approach means satellites collect only the most valuable images, reducing the amount of unusable data returned to Earth. All these actions, from adjusting the satellite to decision-making and updating the imaging plan, happen in a rapid 60–90s cycle as the satellite orbits at nearly 17,000mph.

Next Steps: From Cloud Dodging to Crisis Detection

Currently, Dynamic Targeting helps satellites skip cloudy images, but the vision is much broader. Future tests will teach the AI to actively look for significant weather features like storms, thermal anomalies (such as wildfires or volcanic eruptions), and other fleeting phenomena. Each event requires specialized algorithms that quickly recognize and react to different patterns in real time—a leap toward operational science missions that can seize rare observation opportunities.

The Future: Networks of Intelligent Satellites

NASA isn’t stopping with a single smart satellite. Plans are underway to test Federated Autonomous Measurement (FAME), enabling multiple satellites to work as a collaborative team. A lead satellite could analyze incoming images and direct others to focus on targets of interest, allowing the whole constellation to concentrate on critical developments as they happen.

Dynamic Targeting might also be adapted for deep space exploration, as shown by earlier experiments using the ESA Rosetta orbiter to spot emissions from comet 67P/Churyumov-Gerasimenko. On Earth, future radar-equipped satellites could use similar look-ahead sensing to monitor severe weather like deep convective ice storms in real time, supporting NASA’s goal of fielding more agile and responsive measurement systems across a wide range of missions.

Advancements in onboard AI are setting the stage for a new era in Earth observation—from skipping clouds to detecting wildfires and beyond, paving the way for smarter, faster, and more targeted satellite science.

Citations:

Lounis, Melissa Ait. “NASA’s AI Satellite Just Made a Decision Without Humans — in 90 Seconds.” Daily Galaxy, Daily Galaxy, 26 July 2025, https://dailygalaxy.com/2025/07/nasa-ai-satellite-decision-without-humans/. Accessed 26 July 2025.