Evolutionary optimization of distributed part handling control

Explores distributed controller strategies for a cellular part conveyor, which is a planar array of micro-actuators with sensors and control logic. Our approach is based on 2D cellular automata theory. Particular concern is taken over a rule representation that is adequate for the optimization by an evolutionary algorithm. A simulation of the part conveyor, including the part dynamics and a friction model, is used to evaluate the quality (fitness) of the rules. The genetic algorithm finds solutions for the task of simultaneous positioning and alignment of a square-shaped part and evolves coordinated behavior over global distances.