Fruit recognition method for a harvesting robot with RGB-D cameras

Abstract In this study, we present a recognition method for fruit-harvesting robot to automate the harvesting of pears and apples on joint V-shaped trellis. It is necessary recognize three-dimensional position target by insert its end-effector. However, RGB-D (red, green, blue depth) camera has problem in that point cloud obtained outdoor environments can be inaccurate. Therefore, propose an effective fruits using not only information from but also two-dimensional images camera. Furthermore, report determining ripeness fruit detection. Through experiments, confirmed proposed satisfies accuracy required continuously harvest fruits.