In industrial manufacturing, everyone is talking about Industry 4.0: smart factories, sensor data, cloud computing, and machine learning. It’s extremely common however, for machine learning Industrial IoT projects to never make it out of the “Proof of Concept” phase.
In this talk I’ll describe how we used Elixir to help bridge this gap, and the production system we built to go from PLC (programmable logic controller) sensor data on a fleet of machines, to a cloud-based application for machine monitoring and anomaly detection.
Nerves, NervesHub, Phoenix, Broadway, and Ecto form the data pipeline and much of the application back-end, which is rounded out by tools and resources from AWS (AWS IoT Core, SQS, S3, Lambda, and API Gateway), python libraries for machine learning, and a React front-end.
I’ll also discuss team dynamics and the multi-disciplinary approach required to build these types of systems, with Elixir developers able to play a key role in project success.