Vision-language models deployed on satellites for autonomous Earth observation

Arxiv CS.AI (top papers)

Two teams have deployed vision-language models on satellites for onboard inference, reducing downlink bandwidth. A new paper describes NAVI-Orbital's zero-shot VLM, while Loft Orbital's YAM-9 runs Google's Gemma 3 in orbit.

Vision-languageearth-observation

See 1 more source

A satellite is now running Google's Gemma 3 vision-language model in orbit, doing onboard inference instead of downlinking everything first4 hours agoJustgototheeffinmoonDiscuss

Arxiv CS.AI (top papers)·Juan Manuel Delfa VictoriaTaran Cyriac JohnAndrew W. Herson

Jun 18, 4:00 AM