Scientists from the Fraunhofer Institute for Production Technology and Automation IPA, together with project partners, developed an automation solution that removes weeds and at the same time loosens the soil around the plants keep them healthy. This solution has already proved a success in field trials in buxus cultivation. It originated within the framework of the ‘AgriApps’ project, in which the project partners work on Apps-based automation solutions for the field operations.
The manipulator with a tool travels along box shrubs, removes the weeds growing around it and makes the bottom loose. Cheap, environmentally friendly and sustainable. This should apply to an innovative and efficient agricultural technique in countries with high wages, in order to remain competitive. Robust autonomous systems can make a contribution here, by spraying and fertilising the labor-intensive activities selectively, accurately and economically, or by performing mechanical weed control.
“More robots are growing in agriculture”, explains Martin Hägele, head of the robot and assistance systems department at Fraunhofer IPA. “Robot technologies such as autonomous navigation contribute to the automation of agricultural machinery. But robotic systems for special applications in agriculture are often specialized and used for defined use and application. In addition, they are only useful for purposes in a particular season “. That is why the project partner’s goal in ‘AgriApps’ is to make the agriculture robots more modular and freely configurable for general applications.
A platform for many applications
To achieve this, the scientists use the system with the system. App concept. This includes the ‘BoniRob’ platform, which runs autonomously through the cultivated zones on four wheels. Various application modules, abbreviated Apps, can be connected to this platform, which take over activities in agriculture. Depending on the desired application, the App is equipped with various sensors and actuators. By changing the App, the platform can carry out seasonal activities such as harvesting, pest control or weeding all year round. This increases the use of a carrier platform, which is an important requirement for economic efficiency.
Accurate and flexible operating concept
The weed control application module developed in the context of ‘AgriApps’ uses an optical sensor system and image processing software for object recognition. This is based on color, texture and depth information to distinguish between weeds and boxwood shrubs. This is done continuously while driving through the nursery field.
For the mechanical weed control the moving part of the application module, the manipulator, used. This harrows the soil with the help of rotating discs, which are fixed to an arm with three axles. As with the object recognition, it is also not necessary during the operation for the carrier platform to stop. This is possible because the control of the movement of the manipulator at any time aligns the information from the image processing as well as the movement of the carrier platform.
In the free space between the box the tool deviates further to the maximum remove weeds. If a box is recognized, the tool is retrieved and is controlled according to the contours of the bush without damaging it.
“This weed control concept makes the weed control agile, fast and efficient”, says Felix Meßmer, project manager at Fraunhofer IPA, about the most important technical components for the application module. This is based on proven control concepts for mobile platforms from the service robotics such as Care-O-bot or unmanned transport vehicles for internal logistics.
The scientists have adapted these control concepts for application in agriculture. This makes possible the reactive movements between platform and manipulator, as are also known from the agricultural machinery in the “Tractor-Implement-Management Systems” and can compensate or support them. Navigates the platform to stay in the track to the right and then moves away from the box shrub, the manipulator responds by sending the tool further out, so that it can be better with the weeds.
The next step is testing the App planned during continuous operation. The technologies that emerge from this development could be used for more applications in agriculture, depending on market needs. In addition, the synchronized control concept can also be used for industrial applications such as mobile production assistants.
The Bundesministerium für Bildung und Forschung subsidizes the collaborative project with project partners InMach Intelligent Maschinen GmbH, Robert Bosch Start-up GmbH, Deepfield Robotics, Fraunhofer IOSB and Fraunhofer IPA.