Archive — US — techcrunch
A satellite just learned to find things on its own — here’s what that means
For the first time, an Earth observation satellite, Yam-9, has autonomously identified areas of interest using a vision-language model (VLM) aboard the spacecraft, indicating a significant shift in how AI could enhance satellite capabilities by reducing data overload and enabling real-time monitoring.
Key points
- Yam-9 identified targets independently using Google's Gemma 3 VLM, improving efficiency in data analysis.
- The use of onboard AI reduces the need for extensive human data processing on Earth.
- Long-term implications include the potential for continuous monitoring and direct communication with satellites.
- Loft Orbital's business model focuses on providing satellite deployment as a service, exemplified by a deal with EarthDaily to operate six satellites.
- Lessons learned from deploying smaller models will inform future applications of larger-scale computing in orbit.
Why it matters
- This milestone represents a trend towards autonomy in satellite operations, aligning with 2025 IT trends emphasizing edge computing.
- The use of VLMs signals a shift towards more intelligent, capable AI applications in space technology.
- Ongoing research at companies like Planet Labs and Kepler Communications suggests a growing interest in AI-enhanced space analytics.
The successful implementation of VLMs in space represents a transformative step in satellite technology, enabling real-time analysis, automated data curation, and paving the way for future AI-enhanced scientific exploration.