The research reported herein is an empirical investigation of the capability of an artificial neural network to determine part pose by processing image data from the silhouette of a back-lit part. The chief potential benefit of this new approach is simplicity of training, which is important for flexible automated parts feeders. While this research is of general interest to the designers and users of vision-based, flexible parts feeders, these experiments focused on a flexible tray loader that has been developed at the University of Washington. Pertinent aspects of the hardware and software for that system are presented. A relatively simple neural network is proposed and evaluated experimentally. For five of the six challenging parts considered, the network performed flawlessly. The difficulties encountered with the sixth part have been identified, and further work exploring solutions to the problem are underway.

This content is only available via PDF.
You do not currently have access to this content.