Automating CAM with AI: lessons from applying deep learning to geometry
Recorded at CDFAM Computational Design Symposium, NYC, 2024
Full Presentation: th-cam.com/video/VhZtN34vZ1w/w-d-xo.html
cdfam.com/24-nyc/
Organization:
Oqcam
oqcam.com/
Presenter:
Ben Schrauwen
Presentation Abstract
Computer-Aided Manufacturing (CAM) has revolutionized the manufacturing industry over the past century by enabling the use of software tools to generate machine programs. However, a significant limitation remains: these tools still require substantial input from highly skilled human operators. As production technologies have advanced - from multi-degree-of-freedom (multi-DOF) robots to 3D printers and complex milling machines - the complexity of programming these machines has also increased. This growing complexity has made CAM a bottleneck in the adoption of advanced production techniques, particularly as batch sizes shrink and CAM-associated labor costs per part rise.
At two companies I am involved with: ArcNC, where we focus on CAM for robotic welding, and Oqcam, which specializes in dental CAM; we have explored various deep learning techniques to automate different aspects of the CAM process. In this talk, I will provide a high-level overview of our approaches, share key learnings from our journey, and discuss potential future directions for integrating modern deep learning approaches into CAM and design.
Speaker Bio
Ben Schrauwen is an investor and entrepreneur, currently the Co-Founder of ArcNC, Oqcam, and Raidyn. He previously co-founded and served as CEO of Oqton, which was acquired by 3D Systems. Before that, he was a Senior Director in Autodesk’s manufacturing division. Ben also served as a Professor at Ghent University, where he founded a pioneering machine learning research group. He holds a PhD in Computer Engineering from Ghent University and was a visiting researcher at Harvard.