Robot for harvesting cauliflower, and the cutting of cauliflowers

Abstract

The human arm bounded by soft tissues and muscles is capable of fast movement with high precision fidelity. And it is soft as it has muscles and other tissue. Nowadays, robots can do many human tasks. Robot arms are also becoming softer, to make them stronger and safer to use around humans while working in real-world environments. In this study, the focus is on a robot for harvesting cauliflower, and the cutting of cauliflowers in particular. The robotic platform is designed to reuse modular robotic components from other crops and/or different cauliflower varieties. The platform has two robot arms with variable-stiffness technology. The first arm is for cutting the cauliflower in its steam. And the second is for picking the cauliflower. The GummiArm is a 7+1 DOF robot arm and is an open-source project. Here it is used with a cauliflower-specific end effector, which is a cutter designed with 3d printer, while the second arm has a gripping end-effector. The bi-manual configuration allows the separation of grasping and cutting behaviours into separate robot manipulators, enabling flexibility to adapt to different varieties. Here the focus was on the cutting, and on the control of these through the Robot Operating System (ROS). Several experiments were performed with a force-Analysis of the cutting behaviour, during teleoperation, and when using a control exploiting the passive compliance of the GummiArm. These early experiments with the laboratory platform demonstrate the platform's promise, but also a set of challenges to tackle. This new data can be used to compare human labour performance, develop operational concepts and business plans, and drive future design decisions.