Open Access Dissertation
Robot vision; Factories--Lighting;
Industries look toward computer vision as a means to automate materials handling. To make this choice more appealing, useful and feasible vision applications must be explored. However, illuminance variation in the factory environment can undermine the capability and applicability of vision-based control systems. The purposes of this study were (a) to design and develop a vision-based robot material sorting system, (b) to determine the optimal settings for this system under fluorescent and incandescent lighting for two different color parts on a moving conveyor, and (c) to determine the sorting performance of this system under each light source.
The main components of this experimental system consisted of: (a) a SCORBOT ER-V plus (ESHED ROBOTEC) robot system with a slide base and a speed controlled conveyor, (b) a ROBOTVISIONplus vision system, (c) an incandescent lamp light intensity controller, and (d) two PCs. By integrating these components, color sorting applications were developed.
This study was limited to the sorting of two parts with different colors and similar geometry from a moving conveyor. The system performance (sorting) data were collected on the developed application under fluorescent and incandescent light sources with a conveyor speed of 50 mm/s at various illuminations.
This study explored two sorting methods. Method A used the difference in object descriptors to separate the dominos. Method A worked in a limited range of illuminance and identification tolerance for both light sources. Method B used the difference in the observed “saturation” response of the charge coupled device camera to difference in the observed “saturation” response of the charge coupled device camera to separate the dominos. The “saturation” response region described the interval of illuminance where changes in illuminance did not produce a corresponding change in objects’ measured grayvalue. Method B worked in a wide range of illumination with no stipulation on identification tolerances for both light sources.
Unexpectedly, the light sources in the laboratory setting exhibited a large amount of illuminance variation, and these variations caused applications developed using Method A to perform erratically for both light sources. However, Method B sorted reliably over a wide illuminance range: 150 to 1500 lux for incandescent and 214 to 760 lux for fluorescent.
Year of Submission
Doctor of Industrial Technology
Department of Industrial Technology
Ahmed Elsawy, Advisor
John Fecik, Co-Advisor
1 PDF file (x, 126 pages)
©1996 Johnny T. Li
Li, Johnny T., "Effect of light source on the sorting performance of a vision-based robot system" (1996). Dissertations and Theses @ UNI. 912.