Automatic Part Program Generation Essay

972 Words Apr 12th, 2013 4 Pages
An Expert System for Industrial Machine Vision
Yoshifumi KITMURA, Hiroaki SATO, and Hideyuki TAMURA
Information Systems Research Center, Canon Inc. Kashimada, Saiwai-ku, Kawasaki 21 1, Japan

Abstract An expert system for vision algorithm design is presented. Its knowledge-base includes human experts' knowledge about image processing techniques, and is capable of solving given vision problems. As a problem domain, we choose vision algorithms for a parts-feeder, which determines the attitude of mechanical parts on a conveyor-belt and rejects parts with inappropriate attitudes. The expert system for parts feeder is designed to consist of three components: FSE (Feature selection expert), IPE (Image processing expert), DTG (Decision tree
…show more content…
Thus, in every case we must find the possible solutions under the given conditions. The expert system is capable of designing such vision systems and their algorithms. As a practical task, we consider vision algorithms for a parts-feeder. They determine attitude of mechanical parts on a conveyor-belt and reject the inappropriate ones to assembly machines. The expert system may assist the measurements of shape features of parts and generates the decision tree of object recognition. It is assumed that the expert system may help a production engineer finding the solution when the similar tasks and images as ever are given. The idea of this expert system was proposed in [6]. In this paper, the detail of developed system is described, emphasizing the knowledge to select effective shape features and the framework of its representation.

2. The problem : Determining the attitude of parts for an industrial parts-feeder In order to assemble mechanical parts with an industrial robot, it is expected that a few attitudes out of several stable states of the same part are accepted by the system. Hence we use the machine vision system shown in Fig.1, in which all the parts on the conveyor-belt are observed by a TV camera, and then the inappropriate ones are rejected through image processing operations and decision tree, which determines the attitude by comparing measurements of shape feature values with a threshold value. Here

Related Documents