Robotic Device for Identifying and Picking Apples

Abstract

The article presents a device for identifying and picking apples. The design of the device and the fruit picking control system have been developed. Device characteristics: arm reach without gripping-1400 mm, with gripping-1490 mm, load capacity-500 grams, movement speed-25 degrees per second, rotation on the axis of 270 degrees, the number of degrees of freedom 3 with the possibility of increasing the forearm reach, used controller Stm32 nucleo-f207ZG. A deep learning neural network has been developed to identify apple fruits on the obtained camera frames. The created algorithms for processing the input graphic data and the operation of the computer vision system allow you to recognize apple with a high degree of accuracy up to 88.39%.