Fish Catching by Robot Using Gazing GA Visual Servoing.

Abstract

This paper presents a new method of scene recognition for manipulator real-time visual servoing, which utilizes a hybrid genetic algorithm (GA) in combination with a model shaping a target of known shape, and the unprocessed gray-scale image of a scene. The scene recognition method presented here is concerned with the simultaneous recognition of the shape and detection of the position and orientation in the two-dimensional raw-image, of a three-dimensional target being imaged. The proposed hybrid GA employs the "global" search feature of a two-point crossover of a GA, to search a target, together with a GA-based "local" search that focuses on the target of interest found so far, in order to detect accurate target's position in a short time by intensive searching. In order to appraise the proposed hybrid GA recognition method, experiments to pick up a natural fish swimming in a pool by hand net of a robot manipulator by using the visual servoing, have been conducted to show the performances with respect to recognition accuracy in time response and the real-time feature.