Development tools for a model directed workpiece recognition system

作者:

Highlights:

摘要

The application of industrial robots for automation can be greatly extended if the robot is equipped with a visual sensor that allows it to recognize and locate parts and workpieces. Structural recognition methods are frequently used for the recognition of industrial parts and the recognition is guided by an abstract description (the model) of the workpiece. In an industrial environment not specially trained personnel has to be able to establish the model effectively and easily. The paper describes tools that were developed to generate and optimize the model data interactively, which are necessary to drive workpiece recognition systems. The power of these tools for interactive model generation is demonstrated by an example of a typical industrial recognition task. The tools can be further improved towards automatic model generation.

论文关键词:Pattern recognition,Structural pattern recognition,Model directed pattern recognition,Evolution strategy,Workpiece recognition,Overlapping workpieces,Model generation,Model optimization

论文评审过程:Received 28 June 1985, Accepted 12 November 1985, Available online 19 May 2003.

论文官网地址:https://doi.org/10.1016/0031-3203(86)90052-X