On-orbit edge computing solutions


Hardware and AI software payload for on-orbit image processing

SV Image_4-cutout.png

Flexible machine vision software running custom-trained image classification and object detection algorithms

 Compact and powerful data processing unit for complex image analysis on-orbit

Exo-Space's FeatherEdge Platform provides real-time insight on Earth Observation data
  • Each day satellites send over 1 million raw photos to Earth. Images are being created faster than they can be sent to the ground and in the next 3 years satellite cameras will be producing images at a rate of 1 gigabyte per second. That's the equivalent of 400 HD photos every second. Satellites waste approximately 30% of  bandwidth on photos that are unusable (i.e. contain clouds, out of focus etc.)

  • Earth Observation systems are unable to downlink data to centralized cloud services continuously, but they still have a continuous stream of data which requires large amounts of machine learning/compute resources. 


  • Most off-grid worksites have limited bandwidth and unreliable network connections and some data (i.e. for natural disaster response) is only useful if it's processed in real-time.