A satellite just learned to find things on its own — here’s what that means
This title could be clearer and more informative.Try out Clickbait Shieldfor free (5 uses left this month).
For the first time, an Earth observation satellite autonomously identified areas of interest using a vision-language model (VLM) in orbit. Onboard the Yam-9 spacecraft operated by Loft Orbital, NASA JPL's NAVI-Orbital software ran Google DeepMind's Gemma 3 VLM to respond to natural language queries about sensor imagery — classifying land use and identifying infrastructure without ground-based analysts. The milestone demonstrates on-orbit AI data triage, potentially reducing the volume of raw data sent to Earth. Longer term, it points toward larger AI compute infrastructure in space, real-time global monitoring constellations, and AI assistants for astronauts on the Moon or Mars. Planet Labs and Kepler Communications are also exploring similar capabilities.
