BENCHMARK RESULTS AND QUICK FACTS
Machine vision model runs inferences on satellite images in ~3 milliseconds with 96% accuracy
Model is trained to recognize 50 distinct features ranging from mountains to ships to clouds etc.
Capable of processing data in bursts or continuously with a peak power draw of 15 Watts
Dimensions: 97mm x 97mm x 50mm
Mass: 1.4 kg
If more processing is required the system can scale by pipelining models across multiple processors
EARLY WARNING SYSTEM
NATURAL RESOURCES AND MINING
EDGE COMPUTING IN SPACE IS THE FUTURE
With the advent of 5G networks and the explosion of IoT devices and swarm satellites, traditional cloud computing networks will become inadequate and incapable of addressing the growing computational and latency requirements. While edge computing solutions are being adopted for terrestrial applications, there is no established edge computing platform available for space. The space industry currently relies on centralized cloud computing networks on the ground for data processing applications and could hugely benefit from adopting edge computing solutions.