Metrology
We work inside real CMM workflows — programs, fixtures, tolerances, and shop-floor constraints — not a lab bench that ignores how inspection actually runs.
Company
Measurement that can run itself.
Optic Fringe exists so precision manufacturers can inspect parts as automatically as they machine them — without throwing away the CMMs they already have.
Who we are
A deep-tech team at the edge of vision and metrology.
Optic Fringe Corp. is based in Massachusetts. We build software and vision systems that sit on coordinate measuring machines and take over the slow, error-prone steps: recognizing the part, choosing the program, and starting the measurement.
Most machined parts are textureless. The useful signal is in edges and geometry, not surface pattern. That is a hard computer-vision problem, and it is the one we are built around. Solve identification reliably, and a standard CMM can behave like an autonomous cell.
The same stack has a path into CNC tending, robotics, and assembly — anywhere a camera needs to know which textureless part it is looking at.
Financial backing
NSF Small Business Innovation Research
Our work on AI-aided part identification for CMMs is supported by the U.S. National Science Foundation SBIR program (Phase I and Phase II). That research is aimed at a working autonomous measurement workflow: robust recognition under shop lighting, automated routine selection, and feedback when a cycle looks wrong.
We collaborate toward integration with CMM builders and pilot customers across real part geometries — not only lab fixtures.
Why it matters
Few teams know metrology, machine vision, and AI together.
Shops do not need another generic vision demo or another CMM vendor brochure. They need people who understand measurement practice and how to teach a camera to read textureless parts.
Machine vision
Machined metal rarely offers texture. Our vision stack leans on edges, silhouette, and geometry so identity holds under shop lighting and placement variation.
AI
Models bridge live camera frames and CAD-generated views so the cell can name the part, pick the routine, and start measuring with far less manual setup.