Flexible part-feeding system for machine loading and assembly. Part I. A state-of-the-art survey

The cost to feed parts to a robot for either machine loading or assembly in a flexible manufacturing system (FMS) has often been underestimated. A recent survey on part-feeding methods applicable to a board class of flexible manufacturing systems has indicated that part-feeding may comprise two-thirds of the overall investment and are usually the source of a large percentage of work stoppage and defects. The lack of an off-the-shelf data-driven software-control generic part-presenter satisfying both cost and functional requirements is a major source of problems which prevent the flexibility of the overall flexible manufacturing automation to be fully exploited. In this article, the state-of-the-art of part-presentation techniques for flexible part-feeding is reviewed. Special attention has been placed on the use of computer vision techniques which has potentials in adding considerably to flexibility by simplifying grippers, component feeders, and location tooling, and by reducing engineering time required to implement it. The use of computer vision in real-time part-presentation is evaluated in terms of its on-line computational requirements. The evaluation was performed on an experimental prototype based on a breadboard configuration using off-the-shelf hardware. Factors that limit standard machine vision techniques for part feeding are addressed in terms of their reliability