Advancing machine learning for industrial automation at Archronix

Dushyant Puri, PMP, SMART Centre, Anasuya Bhima, David Espinosa Carrillo, School of Applied Computer Science & IT 

Archronix sought to enhance the capabilities of its industrial control unit by developing advanced machine learning functionalities for automotive feature recognition. The existing control unit had the potential to integrate TensorFlow-driven image processing through third-party accelerators. The goal was to create a flexible hardware and software template, leveraging Google Coral and Intel Movidius modules, to optimize machine learning acceleration and performance. 

In collaboration with Conestoga’s SMART Centre, Archronix is focusing on vehicle color identification as an initial application. This four-month project developed a prototype algorithm to demonstrate the image recognition capabilities of Archronix’s hardware platform and assess its compatibility with TensorFlow. By generating a proof-of-concept, this project not only tests the integration of advanced ML capabilities into industrial control units but also sets the stage for future developments in automotive and industrial automation. 

We acknowledge the support of the Natural Sciences and Engineering Research Council of Canada (NSERC).