Target location method Based on Neural Network for eggplant Picking Robot

For the sake of overcoming the shortcoming of binocular stereo vision method such as algorithm complexity and big computational burden, a binocular stereovision method for locating target based on neural network was developed for eggplant picking robot. The G-B gray level image was segmented by means of threshold segmentation on account of the brightness. To meet the vision requirement of the picking robot, the object´s characters, such as outline, area, center of mass, enclosing rectangle and the point to cut off, are distilled. A three layers BP neural network was structured to locate the eggplant. The input variables of the neural network were image center coordinates obtained by two cameras and the output were space coordinates of picking point. The improved BP algorithm was used to train the parameter of the neural network. The effective parameter was achieved after 182 times of training. Experiments showed that the average rms error of the space coordinates of eggplant was 14.7mm and the average time consumed was 0.96s. The target location method Based on Neural Network for eggplant Picking Robot owns good intelligence and wide adaptability and it can meet the demands of the eggplant picking robots.