Fodder beet is a sought-after coarse feed because it results in less nitrogen leaching and significantly more feed units pr ha than maize. The beets can also affect the fatty acid composition of milk in a favourable direction, and more beets in the coarse feed will reduce the currently high amount of starch, primarily from grain and maize. However, more eco-beets will require the automation of both weed control and the correspondingly heavy-duty thinning of beets. Today, more beet seeds than necessary for the target beet stand are sown to compensate for the loss of beet plants caused by diseases, poor seed germination and insect attacks.

The purpose of CROPCUT is to develop a technology for automating the thinning task, which can also further the cultivation of sugar beet for an increasing eco-sugar market.

The purpose of CROPCUT

The purpose of CROPCUT is to develop a new technology for automating the thinning of beet plants. Today, thinning of beets is done manually, and is thus a barrier for the cultivation of more eco-beets. The company Frank Poulsen Engineering participates in the project with the machine Robovator, which is developed for automatic weed control in the beet rows. CROPCUT will expand the machine’s functions with the ability to thin beet stands automatically. An improvement that can increase current acreages with fodder and sugar beet. The technology is also expected to also have great potential for thinning out of sown vegetables.

The project step by step

During a four-year period, CROPCUT will:

• Improve software for recognition of beet plants based on artificial intelligence and machine learning
• Make decision-making algorithms for automatic thinning of beet stands
• Optimise beet plant populations that unite the demand of a high yield with the possibility to have an effective and automated thinning and weed control
• Improve the weed control robot Robovator, from Frank Poulsen Engineering, with new functions for automated thinning of beet stands
• Evaluate and present the thinning technology under practical conditions in collaboration with ØkologiRådgivning Danmark and an organic milk producer.

Densely sown beets for early thinning. Photo Bo Melander, Aarhus University.

Individual beet plant in a dense weed population – new software technology can distinguish beet plants from weed plants. Photo Bo Melander, Aarhus University.

Identification of beet plants under high weed pressure. With the help of artificial intelligence and machine learning the software can recognise the beets with great probability as shown by the heat map in the bottom row of images; the bigger the red blobs, the more likely it is that it is a beet. The purple dots in the middle row of images show where the software has located the centre of the beet. Photo Frank Poulsen Engineering.

The weed control robot Robovator controlling weeds in weedy eco beets. Photo Frank Poulsen Engineering.

Densely sown beets for early thinning. Photo Bo Melander, Aarhus University.

Individual beet plant in a dense weed population – new software technology can distinguish beet plants from weed plants. Photo Bo Melander, Aarhus University.