Ecological weeding

With the EU-wide ban on plant protection products, agriculture needs alternatives to pesticides in order to still guarantee high yields in the future. Crops compete with weeds for nutrients and this can have a detrimental effect on yields. Mechanical weed removal, particulary using autonomous field robots, is increasingly establishing itself as the method of choice here. A GPS-controlled robot systematically drives along the field and distinguishes crops from weeds using optical sensors and artificial intelligence.  By knowing the exact coordinates of the weed in the field, these plants can then be precisely removed from the ground using mechanical actuators (see figure).

In close cooperation with the seed company Strube D&S GmbH and the agricultural robot developer Naïo Technologies, Fraunhofer IIS has developed the cognitive sensors for the autonomous field robot BlueBob. These cognitive sensors allow BlueBob to detect sugar beets and weeds as it moves along, distinguish between them in real time, and control the robot’s hoeing tools to mechanically remove the weeds.

For this, the sensors are mounted on the front of the robot so that the field and the plants can be captured with high-resolution cameras and forwarded to the AI processor. As the robot moves over the plants, the classification is carried out and the classification result is available when the weeding tools pass over the plants´ location. This means that the weed can be removed efficiently and in an environmentally friendly manner, without damaging the sugar beet. Thanks to precise GPS, plants and weeds can be located to within a centimeter.