Photo: AGCO

Automated harvest

tirsdag 23 okt 18


Ole Ravn
Professor Emeritus
DTU Electro
45 25 35 60

The combine harvester, or simply combine, is a highly complex agricultural machine that performs several different types of tasks simultaneously.

A research collaboration has now made it possible for one of the world’s largest manufacturers to automate this process.

There are four large companies in the world which produce machines for agricultural field work. One of them is American-owned AGCO, which has grown through the acquisition of existing companies worldwide—including the Danish manufacturer of combines, Dronningborg. Part of the company’s development department is in Denmark.

For the past seven or eight years, AGCO has prepared to automate the combine and has therefore contacted researchers at DTU about the development of suitable sensors.

This collaboration accelerated further when AGCO, four or five years ago, decided to create an entirely new global combine platform which is to be used to harvest crops all over the world—from soybeans in South America to wheat in Northern Europe.

The new combine should also be as automated as possible. This requires sophisticated sensors that can perform complex measurements and weightings of the harvested material—and these processes must form an actual automation solution.

"The combine is one of the last agricultural machines to become fully automated because it involves so many processes and sub-systems that require further automation before the operator is completely superfluous. "
Morten Leth Bilde, AGCO

AGCO does not have the necessary expertise to solve the task, explains Morten Leth Bilde, who is responsible for research and development of advanced technology in AGCO.

The company therefore combined forces with DTU on appointing an industrial PhD, Dan Hermann, who quickly became an important part of the efforts to develop the new combine platform.

Small factory

A combine is actually quite the small factory.

Firstly, it harvests the grain in the field, and the grain is then transported through the machine There are several steps along the way that ensure that the grain is separated from straw and chaff. This is done by means of a so-called processor, where the crop is threshed, and the grain is separated from the straw. Then, two sieves ensure that the grain is cleaned and separated from debris. Finally, straw and chaff are blown out so they end up on the ground, while the grain is collected in the grain tank and later unloaded in a grain cart which transports it away from the field.

It is a complex process affected by many factors—for instance, whether the field is uneven or hilly or if the grain is moist from morning dew. Harvesting today therefore requires a number of manual settings of the harvester’s functions, and the driver needs to leave the driver’s cab several times a day in order to check and adjust these.

“The aim of AGCO is to create such as high degree of automation that the combine itself combined with various sensors can measure and act on changes it comes across in the field. The machine must be able to adjust so it runs as optimally as possible and produces the best quality of harvested crop,” says Dan Hermann.

Automation based on data and models

Dan Hermann began by gathering the data necessary for building models of the harvester’s many different work processes. The models are both virtual and physical, and they make it possible to understand how the overall system works—and how the individual work processes in the harvester are weighted and affect each other. Only when the flow through the combine had been described and estimated did the development of necessary software and the advanced sensors begin.

Along the way, it was possible to go back and adjust based on the model’s knowledge of how the processes work individually and in interaction with each other.

Foto: AGCO

Before the development of sensors and software could begin, the flow through the combine had to be described. Illustration: AGCO.

Goals for optimization

Initially, AGCOs new combine has had the first modules for automatizing the process installed, and more will follow later.

Since the functions of the combine’s processes are closely linked, it is not possible to optimize all aspects at once.

Higher speed for instance increases capacity, so many tonnes can be harvested per hour, but the purity of the harvested material will be worsened. If a high degree of purity is important, it will result in larger waste in turn—that is, more grain will end up on the ground alongside the straw and chaff.

Dan Hermann explains that the driver of the combine must determine the optimization aims, after which the machine automatically adjusts to this.

“Our automated combine has default settings which ensure that even inexperienced drivers get a reasonable harvest. Experienced drivers will still be able to manage the settings and choose which parameters to optimize,” he says.

There is five closely connected settings of the combine which AGCO has either automated or which will be automated in the near future. These are the number of revolutions of the rotor, the number of revolutions of the fan, the setting of the top sieve and the bottom sieve for the cleaning process, and the driving speed.

Dan Hermann is currently employed by AGCO, where he is working to further develop and refine the automation of the company’s next generation of combines. The aim is a fully autonomous machine, but it will be a while before we will see this.

“The combine is one of the last agricultural machines to become fully automated because it involves so many processes and sub-systems that require further automation before the operator is completely superfluous. Combines are also very sensitive to external conditions and the nature of the crops,” says Morten Leth Bilde.

He believes that during the next five to ten years, we will see the first fully autonomous machines in the fields—initially, this will probably be for simple operations such as soil cultivation and transport.

Industrial PhD scheme

An industrial PhD is a three-year business-oriented research project and PhD degree programme which is carried out in collaboration between a company, an industrial PhD candidate, and a university.

An industrial PhD student is employed in the private company and enrolled at the University.

The company applies for funding for the project from Innovation Fund Denmark, and the student is employed with pay during the entire project.

Students divide their working hours between the company and the University and spend all their time on the project.

DTU will assist by helping to prepare the application for an industrial PhD project, find suitable candidates, etc.