1990 年 56 巻 529 号 p. 2402-2409
This paper deals with a recognition method of plant devices and equipment using the visual method for the plant maintenance robot. For this purpose, the fiber grating # (FG) method is employed in this paper to identify and recognize three-dimensional plant objects, such as pipelines, flanges and others, which generally have curved surfaces. The FG itself is made of optical fibers through which the laser beams make discrete dot patterns without any mechanical scanning structures, unlike other scanning methods. Thus the proposed method has the following advantages : (1) # it is possible to obtain 3D image information by means of a single camera ; (2) the processing time is generally shorter than that of other methods. Although the proposed method does not give a fine analysis method of image data like common scene analysis, it is suitable to obtain rough information on the working environment of robots. This paper introduces a rule-based expert system capable of recognize plant equipments from the processed image data. The rules are made on the basis of the typical data of each device, irrespective of their sizes. Some experimental results are also shown here to verify the proposed method.