A genetic algorithm for optimizing vector-based paths of industrial manipulators

Nowadays there is a vast amount of IT tools specialized in vector graphics. The data generated by those tools could be used to describe the path of industrial manipulators as a set of vectors. The main problem is that the sequence/direction of those vectors is not meant to be executed by a robot and attempting to do it, would result in inefficient cycle times of the robot. Therefore it is necessary to generate an execution plan that minimizes the cost of carrying out the vector-based path. The number of possible execution actions has a factorial growth and it is unfeasible to evaluate each of them. This paper proposes the use of a genetic algorithm to optimize this task. The main contribution of this work is a chromosome encoding structure and modifications to the Partially Mapped Crossover operator in order to comply with the constraints of this optimization problem. The algorithm was implemented and tested in a real industrial manipulator.