An Intelligent Robotic Platform for Fruit Selective Harvesting

This paper introduces a robotic platform designed specifically for selective fruit harvesting. The platform combines (i) an intelligent module for detecting fruit and classifying its ripeness using multispectral sensing, and (ii) a DMP-based motion planning system that enables a robotic manipulator to pick fruits and place them into a basket. We tested this platform using the Tiago robot, which was equipped with a 7-DoF robotic arm, an RGB-D camera, and a VIS-NIR multispectral device. The robot was programmed to identify six stages of tomato ripeness and to perform harvesting tasks at nine different positions. The results confirmed the platform effectiveness in assessing various ripeness levels, with an average classification accuracy of 93.72%, and its ability to adapt its movements to different targets, achieving a maximum position and orientation error of 8 mm and 0.09 rad, respectively. Additionally, the platform successfully performed all tasks, achieving a 100% success rate.