In this country, pickling cucumbers are harvested by hand, usually with the help of cucumber planes - i.e. vehicles with attached wings. The harvesters lie on their stomachs on the agricultural machinery and pick the cucumbers. Due to the laborious and exhausting manual work involved, this form of harvesting is becoming increasingly uneconomical. In addition, the cost per fruit has risen since the introduction of the minimum wage in Germany. Many German growing regions are therefore threatened with extinction: cucumber cultivation is already shifting to Eastern Europe and India. Improved harvesting technologies are therefore urgently needed to keep cucumber cultivation in Germany profitable.

Together with scientists from Spain and Germany, experts from the Fraunhofer IPK in Berlin are investigating the automation potential of cucumber harvesting in the EU project CATCH - short for 'Cucumber Gathering - Green Field Experiments'. Partners in the project are the Leibniz Institute for Agricultural Engineering and Bioeconomy (ATB) and the Spanish CSIC-UPM Center for Automation and Robotics.

The researchers' aim is to develop and test a cost-effective dual-arm robot system made up of lightweight modules that is suitable for automated cucumber harvesting, but can also be used for other agricultural applications. The harvesting robot should be cost-effective, powerful and reliable and be able to recognize cucumbers ready for harvesting even in adverse weather conditions and then pick and place them gently with its two gripper arms.

Modern control methods help to equip the robot with tactile sensitivity and also enable it to adapt to environmental conditions. They also allow it to imitate human movements. One of the aims is to prevent the plants from being damaged or even pulled out of the ground along with their roots. Another requirement is that the automated harvesting assistant must be at least as effective as the pickers - an experienced picker can pick up to 13 cucumbers per minute.

High hit rate

Visual and tactile detection, assessment and evaluation is a major challenge for autonomous systems. It becomes even greater when green objects have to be located in a green environment, as is the case when harvesting cucumbers. In addition, the fruits grow disorderly in the field and are sometimes covered by leaves. Varying light conditions make the task even more difficult.

Multispectral cameras and intelligent image processing should help to locate the cucumbers and guide the robot's gripper arms to the right place.

This task is the responsibility of the Spanish project partner CSIC-UPM. A special camera system ensures that the cucumbers are registered and localized with a hit rate of around 95%. However, the plan is to pick all the ripe cucumbers so as not to hinder the growth of the new, regrowing cucumbers. The Fraunhofer IPK developed the robot arms, each with five degrees of freedom, on the basis of hardware modules from Igus.

Searching for a human role model

The IPK experts' task in the project is also to develop three different gripper prototypes: a gripper based on vacuum technology, a bionic gripper jaw (FinRay) and a 'cucumber hand' modified on the basis of the OpenBionics Robot Hand.

The cucumber harvest of tomorrow? The prototype of the dual-arm robot system during initial field tests.

© Fraunhofer IPK

They are building on work from another European research project in which they had already developed a dual-arm robot controller with efficient task-oriented programming for Workerbot I, a humanoid robot for industrial assembly. This control system is now being expanded for planning, programming and controlling the robot's behavior during cucumber harvesting.

The pre-programmed behavior patterns enable the robot to search for cucumbers bimanually in a human-like manner: "For example, it can push leaves aside using symmetrical and asymmetrical or congruent and incongruent movements. It can also automatically switch on-the-fly movements to approach an identified fruit and then grasp it," says Dr. Dragoljub Surdilovic, a scientist at Fraunhofer IPK. The researchers' goal is an intelligent control system with 'judgment' that distributes the tasks between the gripper arms, monitors the picking process and can handle exceptions.

An initial field test of the robotic system took place in July 2017 on the test field of the Leibniz Institute for Agricultural Engineering and Bioeconomy (ATB) with various cucumber varieties. The harvesting of new varieties was also tested - with characteristics that facilitate automatic recognition. The initial tests have confirmed the basic functions. Since fall 2017, the project partners have been continuing their experiments in a greenhouse. The focus of the investigations is on testing the efficiency and robustness of the system against disturbances. Once the tests have been completed, the lightweight robot will be ready for the market.