Robot programming with a genetic algorithm

Abstract

Robot systems are an archetype of complex process control systems where valid production programs may have any number (>2) of process commands which are difficult to optimize with classical methods. Robot trajectories can be programmed automatically by a genetic algorithm, providing the algorithm will consider the order and varying lengths of trajectories. A genetic, adaptive, heuristic algorithm which is designed to optimize robot trajectories is described. Though the model presented is a general model for redundant structures and could represent any n-link structures, it was applied to a 3-link structure. The performance of the genetic algorithm shows characteristic improvements when compared with that of a hill-climb and random search algorithm.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>