Self-driving weed killer
© Bundesanstalt für Landwirtschaft und Ernährung/Fraunhofer
December 2021

Self-driving weed killer

By Björn Carstens
Most farmers would prefer other options than the massive use of chemicals for weed control. But sometimes there’s just no way around it. Or is there? Researchers from Fraunhofer have now presented an autonomous weeding robot requiring neither sprays nor costly high-tech sensors including complex algorithms.

Weeds are truly a pest, depriving crops of light, water and nutrients. They inhibit crop growth especially in the early stages. For owners of private gardens, that’s a nuisance while the owners of tree nurseries or orchards incur high costs for complex weed control by flaming, grubbing, hoeing or spraying with herbicides that are harmful to the environment.

Fraunhofer Institute for Manufacturing Engineering and Automation IPA in Stuttgart in collaboration with partners has developed an alternative in which a machine performs the time-intensive weeding job. AMU-Bot (AMU stands for “autonomous mechanical weed control” in German) is an electric caterpillar autonomously traveling through plantations and killing weeds by means of rotating blades aka rotary harrows. At the end of a row of trees the weeding robot autonomously turns around and enters the next row.

Self-driving weed killer
The manipulator also moves in the gaps between the crop rows and removes any weeds here using the rotary harrows© Bundesanstalt für Landwirtschaft und Ernährung/Fraunhofer
Navigation with LiDAR scanners

The machine uses optical sensors and GPS for guidance. The installed LiDAR (light detection and ranging) scanners with their continuously emitted laser pulses detect potential obstacles and guide the robot through the plantation with pinpoint accuracy.

A height-adjustable tool uproots the weeds that will subsequently dry up and no longer compete with the cultivated crops and therefore do not have to be collected separately. AMU-Bot removes weeds between the crops within a row by means of laterally extensible circular harrows.

Self-driving weed killer
The autonomous caterpillar vehicle, AMU-Bot, drives between the rows in the tree nursery and removes any weeds© Bundesanstalt für Landwirtschaft und Ernährung/Fraunhofer
Technology for networked farming

The project managers deliberately opted for a seemingly simple solution. “A system that classifies the different individual plants requires high-resolution cameras, AI-supported image recognition algorithms and plant profiles stored in a database. These systems are far more complex and expensive. Not only that, but they cannot readily switch to working in new contexts,” explains Fraunhofer researcher Kevin Bregler.

The AMU-Bot platform removes weeds reliably, is economical, robust, easy to operate and highly efficient.

Fraunhofer researcher Kevin Bregler

When the mobile weed killer prototype will be ready for market is not clear yet. The experts at Fraunhofer Institute view AMU-Bot as a piece of a puzzle in a forward-looking vision called Cognitive Agriculture, where digital services and tools will be collecting data about farmland ultimately leading to “flexible and intelligent automation of sustainable agriculture” – including weed control.

Self-driving weed killer
The LiDAR scanner (red arrow) installed in AMU-Bot continuously emits laser pulses as the vehicle moves, which the system uses to determine the position of the crop rows© Bundesanstalt für Landwirtschaft und Ernährung/Fraunhofer