AbstractFor flexible manufacturing of short production runs where a large variety of product sizes, component types, and surface reflectance characteristics are encountered, it is desirable to build flexible computer‐controlled systems for feeding parts into machine tools or assembly processes that combine maximum flexibility and reliability with minimum cost and cycle‐time. This is not a general bin‐picking problem; the parts are assumed to be propositioned approximately in totes/pallets/kits with regularly spaced locations. This article presents a machine vision technique based on the principle of retroreflective vision sensing for part‐presentation. Since retroreflective material has a distinctive surface reflectance that is not commonly found in natural or man‐made objects, the use of retroreflective surfaces enables reliable high object‐to‐background contrast images to be obtained for a wide variety of objects. Unlike conventional machine vision techniques, which rely on the variance of the surface reflectance of the objects to generate detailed images, retroreflective vision sensing aims at generating a reliable two‐dimensional digital object silhouettes so that the location and orientation of the part can be reliably determined. Two application examples, machine loading and assembly,
展开▼
