Self Organizing Feature Maps and Their Applications to Robotics

Loading...
Thumbnail Image
Penn collection
Technical Reports (CIS)
Degree type
Discipline
Subject
Funder
Grant number
License
Copyright date
Distributor
Related resources
Author
Sayers, Craig
Contributor
Abstract

The self-organizing feature maps developed by Kohonen appear to capture some of the advantages of the natural systems on which they are based. A summary of the operation of this form of artificial neural network is presented. It was concluded that the primary benefits of using self-organizing feature maps result from their adaptability and plasticity while most problems are largely caused by the lack of a rigorous mathematical foundation. Two different robotics applications are described. In the first, developed by Martinez and Schulten, a hierarchical structure composed of many self-organizing feature maps is used to control a five degree of freedom robot arm. While it was noted that there may be some practical problems, the general idea of using a hierarchical structure appears sound and may be applicable to a wider range of problems. The second robotics application was developed by Saxon and Mukherjee. They used a single self-organizing feature map to learn the motion map of a two degree of freedom arm. The use of such a system should simplify path planning by combining multiple constraints into a 2-D structure.

Advisor
Date Range for Data Collection (Start Date)
Date Range for Data Collection (End Date)
Digital Object Identifier
Series name and number
Publication date
1991-05-01
Volume number
Issue number
Publisher
Publisher DOI
Journal Issue
Comments
University of Pennsylvania Department of Computer and Information Science Technical Report No. MS-CIS-91-46.
Recommended citation
Collection