Hoeing is a promising alternative to chemical weed control. Manual steering of hoes can be replaced by automatic steering systems based on GNSS-technology and optical sensor technology.

The department of Weed Science at the University of Hohenheim and the company K.U.L.T. have developed a new camera-steered hoe for cereals which are sown at tight row spacing. The cameras recognize the crop rows even at a width of 13-15 cm and at high driving speed.

The objectives of this project are to optimize and test the new camera-steered hoe in narrow spaced cereals, maize and wide-spaced soybeans and to implement this new technology into practical farming. The following hypotheses are specified:

A: New imaging technology can improve accuracy of crop row recognition and hoeing guidance.
This development is necessary because new sowing technologies arise (e.g. equidistant row spacing) (Consortium Partner 4).


B: The camera can automatically adjust the distance of the hoeing blades.
We will develop a hoe allowing hydraulic horizontal adjustment of the blades from the driver’s cabin in order to receive optimized results without setup time for crops with varied row distances.

C: The new hoeing technology efficiently suppresses weed competition.
Repeated plot experiments and on-farm research studies for a period of three years will be conducted in cereals, maize and soybean at Meiereihof and Heidfeldhof of the University of Hohenheim and in Rheinstetten/Forchheim to test the steering system, weed control efficacy, crop response and efficiency of mechanical weeding and its impact on grain yield and quality.

Setup and conducting plot experiments in summer/winter cereals, maize and soybeans at Meiereihof and Heidfeldhof of the UHOH and Rheinstetten/Forchheim. We test different hoeing blades between and within the crop row, speeds, depths, dates, row distances, crop densities/distributions, sowing technologies, seed-bed preparation and steering technologies. Organization and realization of on-farm experiments in all above stated at 3 additional sites (Hirrlingen, Lindenhof and Stifterhof/Bruchsal).

During the tests weed density will be measured separately for each weed species before and 14 days after hoeing. It will be counted, how many weeds and crops were uprooted, covered and new emerged. It will be determined if hoeing in cereals induces the development of new tillers. Weed control efficacy will be calculated for the inter-row and intra-row area separately during various development stages.

In addition, programing of image analysis software for weed identification and steering of the hoe in order to secure the exact row determination even with overlapping plants in tight crop rows with high weed density. Distinction between “weeds” and “crop plants” will be secured by utilizing artificial intelligence Deep Learning processes.