Sensor technology integration and intelligent decision making in an industrial prototype for herring roe grading

Elizabeth A. Croft, Clarence W. de Silva, Setiawan Kurnianto

Research output: Contribution to conferencePaper


The automated grading of non-uniform products, such as fruit, cuts of meat, and other discrepant goods is an ongoing problem. In some areas technical advances in have permitted the on-line, single-criteria sorting of millions of items per minute. However, for many non-uniform products, especially those for which a multi-factor decision about overall quality is required, grading is still done substantially by hand. Herein, we describe the development and integration of sensor technology for Herring Roe Grading. Size, shape, firmness/texture and weight all play key roles in determining the grade of each roe skein. These variables are subjective both in the way they are determined by humans, and in the way they affect the grade of a skein. Thus, an 'intelligent' decision-making system was used to determine the final grade of the product. The system uses a fuzzy-logic knowledge base developed through acquisition of knowledge from grading experts and off-line experiments.

Original languageEnglish
Number of pages8
Publication statusPublished - 1 Dec 1996
Externally publishedYes
EventInternational Mechanical Engineering Congress & Exposition 1996 - Atlanta, United States of America
Duration: 17 Nov 199622 Nov 1996


ConferenceInternational Mechanical Engineering Congress & Exposition 1996
Country/TerritoryUnited States of America

Cite this