Robotic Green Asparagus Selective Harvesting

Abstract

Robotic harvesting in the open field demands innovative solutions in robot perception and mechanics to cope with environmental challenges. In this paper, a prototype robotic harvester that is able to cope with the challenges of selective harvesting of green asparagus is presented. The harvester is able to drive along an asparagus dam in the field, detect asparagus stalks, identify stalks that are ready for harvesting, and perform harvesting without damaging. These system abilities are enabled by a novel vision perception module and a novel harvesting mechanism. The perception module is based on three-dimensional-point cloud processing, and it is able to reliably and robustly detect the asparagus stalks, their positions, and dimensions. A novelty of the proposed harvesting mechanism is a multi-tools solution to increase harvesting productivity. The results of the first outdoor field tests demonstrate the applicability of the presented robotic harvester.