red 
g aco) 
Light 
  
ght as the 
and 8 show 
resentation 
ults also 
nd 10) and 
les. These 
jecause the 
ches rules 
surface. 
angle. If 
ive to 
  
  
ation 
sition ‘A 
sition ‘B’ 
  
of 
The technique used for the application was very 
similar to the triangulation technique 
suggested by Lighting Advisor. Figure 11 shows 
the two slits of light projected on the case. 
The light sources were fiber optic bundles with 
shaped apertures and lenses. One light 
produces the triangulation effect for the 
position of the raised surface. The second 
light backlit the hole (not a suggestion of the 
expert system). Figure 12 shows the monitor 
image from a camera looking down on the 
transmission case. From this image, the light 
spots were easily measured between with 
conventional analysis techniques. 
    
Figure 11 
Projected slits of light 
  
Figure 12 
Monitor image of slits 
CONCLUSION 
The field testing of the Lighting Advisor 
expert system has just begun. No conclusions 
have been reached with respect to its 
usefulness. However, the preliminary results 
are encouraging. The development of an expert 
system is never really complete since human 
knowledge and expertise are always expanding. 
With respect to Lighting Advisor, there will be 
continuing additions and refinements. 
Applying machine vision still requires a fair 
amount of highly specialized engineering 
expertise in several disciplines. Expert 
systems can help bridge the knowledge gap in 
many areas of expertise, enabling further 
expansion of the machine vision technology base 
and support. Other areas in machine vision 
where the expert system could play a role are 
system troubleshooting, selecting and 
optimizing inspection parameters, and controls 
and machine interface design. The future looks 
promising. 
References and Acknowledgements 
Techknowledge Inc. 
Texas Instruments 
1EEE Video. Conferences on Artificial 
Intelligence 
Expert System Services Ball Corporation 
© 1988 Ball Corporation