NASA is pushing the limits of space technology with Dynamic Targeting, an AI-driven innovation that allows satellites to autonomously choose observation points from orbit in seconds. This technology could revolutionize how spacecraft collect critical data for Earth science.
In a recent test, NASA demonstrated how artificial intelligence could significantly improve the effectiveness of Earth-observing satellites. For the first time, an AI-equipped satellite was able to analyze images, process data, and autonomously determine where to aim its instruments—all within a remarkable 90 seconds, with no human intervention required. This development promises to increase the precision of satellite data.
Dynamic Targeting, a project developed over the past decade at NASA’s Jet Propulsion Laboratory (JPL), aims to make spacecraft smarter in their operations. This first trial, conducted in July, used a commercial satellite to test the technology’s ability to enhance imaging. The system helps spacecraft avoid cloud cover and quickly identify time-sensitive phenomena like wildfires or volcanic eruptions.
Steve Chien, a technical expert at JPL, explained that the concept mimics human-like decision-making. Instead of merely gathering data, the satellite can analyze the information and respond intelligently. The goal is for satellites to recognize critical events, such as a fire, and direct their sensors accordingly.
Improved Science by Avoiding Clouds
In its initial test, Dynamic Targeting focused on an issue that plagues many Earth-observing satellites: clouds. For optical sensors, clouds can block as much as two-thirds of the view. Dynamic Targeting solves this by looking ahead, distinguishing between clouds and clear skies, and optimizing image collection. If a cloud is detected, the satellite skips imaging to avoid wasting valuable data storage.
Ben Smith, a JPL associate, emphasized that this approach would allow researchers to gather more usable data. By skipping unnecessary cloud images, satellites can focus on what’s important, enhancing the overall quality of scientific data.
How Dynamic Targeting Functions
The tests are being conducted on CogniSAT-6, a small CubeSat launched in March 2024. This satellite, developed by Open Cosmos and Ubotica, is equipped with an AI processor to run Dynamic Targeting. Previous tests on the International Space Station in 2022 showed that the AI-driven system could work for space-based remote sensing.
In practice, the satellite tilts to capture ahead-of-orbit imagery, which is then analyzed by Dynamic Targeting’s cloud-detection algorithm. Based on this analysis, the system guides the spacecraft to areas with clear skies, focusing its sensor on the ground below. The entire process takes just 60 to 90 seconds while the spacecraft travels at nearly 17,000 mph.
Future Prospects
With cloud-avoidance functionality now proven, NASA plans further tests on Dynamic Targeting’s ability to detect severe weather, such as storms and wildfires. The system’s adaptability could allow it to be used in various science missions, opening new possibilities for Earth and space research.
This initiative could eventually be scaled for use on spacecraft exploring the solar system. Chien and his team drew inspiration from the Rosetta mission, where data was used to autonomously identify plumes on comet 67P/Churyumov-Gerasimenko. Dynamic Targeting could offer a similar solution for detecting cosmic phenomena.
Looking forward, the use of Dynamic Targeting could extend to radar-equipped satellites studying rare, dangerous weather events like deep convective ice storms. The system’s AI would automatically focus on storm clouds, collecting crucial data during the satellite’s brief flyby.
NASA is also exploring how multiple spacecraft could work together using Dynamic Targeting. A leading satellite could direct a trailing one to focus on specific events. Chien’s team is testing this concept later this year with a program called Federated Autonomous MEasurement.
