Ann: Vorago Technologies Collaboration, page-344

https://www.nasa.gov/ames/ocs/summerseries/2018/michael-lowry

From 2018 and they didn't know about Akida and Intel had just released their Neuromorphic chip

Now we have Akida NSoC being introduced to NASA


Abstract:

After fifty years of exponential improvements in Von Neumann computing, Moore’s Law is running out. What will replace conventional computers as the next driver in the decades ahead? One pathway is brain-inspired computing – using lessons from the brain to inform new architectures that are much more energy efficient, capable of massive parallel processing, and capable of learning in-situ.This lecture describes the emerging field of Neuromorphic computing, and how it will change hardware and software engineering. This field has been energized by the success of deep machine learning, which since 2012 has led to computers achieving human-level performance in image classification, speech understanding, and many other tasks. Neuromorphic computing emulates the brain with layers of silicon artificial neural nets that do inference through the weights of artificial synapses between neurons, and in advanced models does machine learning on the chip in-situ. The trade space for Neuromorphic processors will be described highlighting recent chips including Goggle’s TensorFlow Processor Unit and Intel’s Loihi chip. Software engineering will undergo a paradigm shift in the era of Neuromorphic computing reflecting hardware advances. Biographyr. Michael Lowry is the NASA Senior Scientist for Software Reliability. He has served as NASA’s Lunar Surface Systems Software Architect (2008-2011), software lead in the Toyota unintended acceleration investigation (2010), and DARPA chief scientist for the META project investigating cyber-physical system development principles that are correct-by-design (2012 – 2015). Since 2015, he has led as principle investigator the Autonomy Operating System project (AOS). AOS has developed an open flight software architecture incorporating heterogeneous artificial intelligence reasoning capabilities. The goal of AOS is to achieve an autonomous pilot that can navigate and communicate in the National Airspace System. AOS incorporates NASA core flight software (CFS) as its embedded platform foundation; cFS has been certified on more than two dozen space missions.Lowry has been the Ames lead on NASA’s high-performance space computer, whose goal is to provide a radiation-tolerant CPU that reaches tens of gigaflops performance while using minimal power. His interest in Neuromophic computing is to advance aerospace processors further orders of magnitude to enable smart agents for both space and aviation.