Improving CNC Processes Using Computer Vision to Digitise Physical Objects

Student thesis: Master thesis (including HD thesis)

  • Jakob Andersen
  • Thomas Egon Kærgaard
  • Jesper Ravn-Nielsen
4. term, Computer Science, Master (Master Programme)
When working with CNC machines, there is sometimes a need to replicate an already existing object. In the upholstery industry it is normal to have handmade templates, thus there is no CAD file available to cut with a CNC machine. Recreating objects without the original file can be a difficult and time consuming process that involves tracing an object in a CAD program. Additionally, not all who works with CNC machines are necessarily apt at using CAD programs. We present CopyCut!, a system that allows users to copy existing objects or create new objects from sketches. CopyCut! extracts contours from images captured using a camera mounted above a CNC machine, and converts them into vector format, allowing users to use them in CAD programs or with the CNC machine. CopyCut! thus creates a new process for copying objects and allows for creative ways of using objects and sketches to create new products. CopyCut! differs from other similar work in that it allows much larger work area than those described in the related work, and does not use any specialised equipment. CopyCut! is a simple addition to an industrial CNC machine, meaning it can be set up and used without changing expensive equipment. We also describe several creative ways to use CopyCut! beyond copying of objects. To evaluate CopyCut! we tested the accuracy and conducted both an expert review and a co-design session. The system has an minimum accuracy error of 0.2 millimetres in the centre of the work area, but worse further away due to camera lens distortion. The expert review and co-design session gave us feedback on the usefulness of CopyCut! and possible improvements.
LanguageEnglish
Publication date6 Jun 2018
Number of pages36
ID: 280504799