Hex head nuts and bolts are almost universally the symbol of mechanical engineering, and their presence is everywhere. Unfortunately, every shop has a pile of fasteners that no one wants to sort, and while these fasteners are valuable, they aren’t worth the time to categorize them. Scrap companies also have large amounts of fasteners that are too expensive to sort, so they are simply sold as scrap steel.
Bolterizer is an automated system for categorizing nuts and bolts. The system uses a webcam to see a selection of nuts and bolts, and then identifies the dimensions of every nut and bolt in the image. Once these nuts and bolts are categorized, conceivably they could be passed to a pick-and-place system to physically sort the fasteners. This would be incredibly valuable for a scrap company, since they could pour fasteners into the machine, turning them from scrap metal into sorted fasteners, which is a steep increase in value from approximately $0.015 per pound (Rockaway Recycling, 3/13/16) to $1.80 per pound (at Fazzio’s Surplus, Glassboro NJ). The machine would pay for itself quickly.
Bolterizer uses a Microsoft Kinect as a webcam to capture the image of the nuts and bolts, on top of a sheet of paper. The corners of the sheet of paper are mapped to an 8.5” x 11” image at 60 DPI, which accounts for perspective and paper position. A contour finder is used to find the profile of the bolts, and an algorithm was developed to pick out key nut and bolt properties from the nut or bolt contour. All of this was completed in 450 lines of C++ and runs at approximately 30 Hz single threaded on a Core i5.