Application of a genetic algorithm to an actuation mechanism for robotic vision

This paper deals with the development of a miniature camera actuation device and its associated control system for use in mobile robotic vision applications. Two lightweight cameras for stereo vision are connected to push-pull actuators and a novel multivariable method is used to balance the control signals aimed at producing high acceleration and rapid scene shifting through open-loop actuation. Once determined, the optimal signals are stored and remain constant in structure provided that the trajectories satisfy certain performance specifications within given limits. The emphasis of this paper is on the algorithm used to determine the optimal control signals. The advantages of genetic algorithms over conventional methods are first demonstrated. A genetic algorithm is then applied to a model of the camera actuation system and typical results are presented. The potential of the proposed method is discussed.