An optimal scheduling of pick place operations of a robot-vision-tracking system by using back-propagation and Hamming networks

The authors present a neural network approach to solve the dynamic scheduling problem for pick-place operations of a robot-vision-tracking system. An optimal scheduling problem is formulated to minimize robot processing time without constraint violations. This is a real-time optimization problem which must be repeated for each group of objects. A scheme which uses neural networks to learn the mapping from object pattern space to optimal order space offline and to recall online what has been learned is presented. The idea was implemented in a real system to solve a problem in large commercial dishwashing operations. Experimental results have been shown that with four different objects, time savings of up to 21% are possible over first-come, first-served schemes currently used in industry