Machine Learning Space Applications Using RC64 Rad Hard Manycore Processor

Ramon.Space RC64 rad-hard manycore processor is designed for artificial intelligence/machine learning computations based on deep learning. VGG-based neural network classification of Space imagery of three different types is demonstrated. Machine learning models and imagery are provided by Aiko Space, Mobius Labs and Space Shift. Image frames of 32×32 pixels are classified at 65 frames per second, achieving power efficiency of 12.5 frames per second per Watt.