CubeSats are becoming regular parts of science collecting, not just experiments as they were treated initially. The problem with these satellites is the sheer number of spacecraft generates so much data, and only a fraction can be sent back. 

Researchers at Carnegie Mellon University want to implement a new system where the constellation could do some of its own processing in orbit using machine learning and only send back the most important data to the ground. They call this system “orbital edge computing”. One suggested application for this technology is imaging constellations. One example provided by PI Brandon Lucia is spotting wildfires before they get big enough to be a problem.

A new grant from the National Science Foundation (NSF) will allow the team to fund lots of grad students to work on the project and build some spacecraft prototypes for launch in a few years. Congratulations to the team at Carnegie Mellon. We look forward to hearing more about this new technology.

More Information

CMU press release