Automated Visual Inspection of Moving Custom Parts

Abstract

Abstract-Mass produced custom parts require inspection routines that can facilitate variations in product parameters such as dimensions, tolerances, and throughputs. Quality control and inspection of these parts, and part families, need to occur at higher frequencies than batched produced parts. This higher frequency of inspection significantly impacts inspection times, and inherently, production rates. An effective, diverse, accurate, robust, and time efficient method for inspecting custom parts is therefore needed. Vision systems are a continuously evolving method of quality control and part inspection. These systems offer the potential to be exceptionally diverse and effective in their applications, and are therefore suited to inspecting custom parts. This paper details the research, design, construction and assembly of a prototype apparatus, which provided a suitable environment in which customized parts were inspected. System integration using the Mechatronic Engineering approach was performed to integrate vision, sensor articulation, and control systems. The apparatus was tested in a Computer Integrated Manufacturing (CIM) cell to quantify system performance. Intelligence was incorporated into the inspection routine by performing visual inspection of only significant Regions of Interest (ROI). Dynamic access by the vision sensor to the various ROIs, allowed for inspection of moving parts, which lead to an increased process efficiency. The eliminated stoppage time required by typical inspection routines, allowed for preservation of specified production rates whilst increasing frequency of inspection.

Machine parts recognition and defect detection in automated assembly systems using computer vision techniques

“…A non-contact automated inspection system to detect defects in moving custom parts was proposed by (Shaniel Davrajh et al, 2008). They have inspected defects by significant Region Of Interest (ROI) feature along with Discrete Fourier Transform (DFT).…”