GEOMETRYLAB • PART RECOGNITION • MANUFACTURING SIGNALS
Turn CAD-derived geometry into recognizable engineering context.
GeometryLab analyzes geometric signals from parts and assemblies to help recognize part types, manufacturing resemblance, patterns, thickness behavior, and engineering intent.
It is designed to sit between CADSync and DEP: CADSync captures the data, GeometryLab interprets the geometry, and DEP preserves the validated knowledge.
WHY GEOMETRYLAB EXISTS
CAD geometry contains signals that teams often interpret manually.
Designers and engineers look at parts and instinctively recognize plates, brackets, shafts, weldments, machined parts, purchased components, and structural members. GeometryLab is being built to capture those recognition signals in a structured way.
What kind of part is this?
Identify likely part categories such as flat plates, machined parts, structural members, brackets, shafts, weldments, or purchased components.
How might it be made?
Evaluate manufacturing resemblance using geometry signals like thickness, hole patterns, cylindrical faces, fill ratio, body count, and complexity.
Where does it fit?
Connect part-level recognition to assembly context, BOM structure, CAD hierarchy, and downstream knowledge capture.
RECOGNITION SIGNALS
GeometryLab looks for the measurable clues behind engineering recognition.
SHAPE
Part form
Bounding boxes, proportions, volume signals, planar faces, cylindrical faces, and fill ratio help describe the basic shape of a part.
FEATURES
Holes and patterns
Hole counts, bore signals, circular patterns, rectangular patterns, counterbores, countersinks, and tapped-hole signals help describe function.
MATERIAL BEHAVIOR
Thickness and bodies
Thickness signals, body count, shell-like behavior, plate behavior, and complexity help separate flat, machined, structural, and assembly-like forms.
ASSEMBLY
Context
Assembly hierarchy, repeated components, vendor parts, BOM role, and parent relationships help explain why a part exists.
MANUFACTURING
Resemblance
GeometryLab can score resemblance to practical manufacturing categories such as flat plate, machined part, weldment, structural member, or stock component.
VALIDATION
Human correction
Recognition results are meant to be confirmed, corrected, and preserved so DEP can learn the company-specific meaning behind geometry.
RELEASE 1 FOCUS
Practical part recognition before advanced automation.
GeometryLab is not intended to magically understand every design. Release 1 focuses on useful, explainable recognition signals that help engineering teams review, classify, and preserve part knowledge faster.
Explainable scoring
Recognition should show why a part resembles a category instead of returning a hidden black-box answer.
Human confirmation
Users can confirm or correct the analysis so the result becomes trusted engineering knowledge.
DEP-ready output
GeometryLab results can be passed into DEP as validated context for future search, review, rules, and lessons learned.
HOW GEOMETRYLAB FITS
GeometryLab connects CAD capture to knowledge preservation.
CADSync → GeometryLab
CADSync captures geometry-derived properties, assembly structure, BOM data, file relationships, and package context.
GeometryLab → DEP
GeometryLab converts shape and feature signals into part recognition, manufacturing resemblance, and explainable engineering context.
DEP → Future decisions
DEP preserves validated recognition results so future engineers can search, compare, review, and learn from prior work.
MULTI-CAD DIRECTION
Built for CAD-derived data, not one CAD vendor.
GeometryLab is intended to analyze normalized geometry signals produced by CADSync connectors. That allows the recognition layer to grow beyond a single CAD system as CADSync expands from Inventor toward SolidWorks and other platforms.
GeometryLab helps teams understand what the geometry is trying to tell them.
Capture geometry signals, recognize part intent, and preserve validated engineering context inside the Stratic Systems knowledge ecosystem.