Toward a sustainable future with data-driven dairy farming

Maassluis-based Lely was one of the pioneering companies in the development of automatic milking robots. The Astronaut machine, introduced in the early 1990s, was a groundbreaking invention that changed the lives of thousands of farmers. While the company initially also had a presence in forage machinery, this division was sold in 2016, allowing Lely to focus solely on automating dairy farming processes such as milking, feeding and cleaning. Currently, it has a fleet of 90,000 machines operating worldwide.

“The essence of automation in a dairy farm is to give cows freedom with as little interventions from the farmer as possible. We call that ‘free cow traffic.’ Automation also gives farmers more time to focus on tasks that they like and consider important,” notes Lars Hoogweg, Head of Management Software at Lely. He’s responsible for the team that develops Horizon, the company’s farm management solution. Before automatic milking machines, farmers would spend hours a day milking cows by hand, leaving little time for other essential responsibilities like caring for sick animals. Thanks to Lely’s software managing the milking machines, farmers now receive notifications when there might be an issue with a cow’s health. “This allows them to promptly investigate and address any concerns,” adds Hoogweg.

Thijs Voogt, Head of Data and AI at Lely, highlights that the company was an early adopter of software and data as crucial components of its vision of the farm of the future. “Farmers are now able to remotely manage their cows instead of having to run around all day. As a result, their lives have changed tremendously: they can go on vacation and celebrate birthdays, while still maintaining care for their livestock. Their cows also have a lot more freedom: they decide when they want to be milked. I find great motivation in knowing that I have a positive impact not only on the lives of these farmers but also on the overall sustainability of the agricultural sector.”

Data, lots of data

Data plays a pivotal role in Lely’s operations, with their machines collecting vast amounts of data. The milking robots, for example, record important details such as the timing of cow visits, milk production quantities and milk composition. All this data is transmitted to a central PC on the farm, running Lely’s Horizon software. Acknowledging that some rural areas may not have reliable internet connections, Hoogweg notes that Horizon is designed to be able to run fully standalone. “Even without an internet connection, the farm should keep running perfectly.” The cloud also plays a key role in Lely’s solutions, though.

Through the Horizon web application on the central PC, farmers can adjust machine settings via a ‘farmer’s cockpit’ and gain insights into milk production volumes, their margins on individual cows and any irregularities in milk production. Additionally, selected data is sent to the cloud for analysis by machine learning models. This includes sensor data of the machines but also animal data and other external data. Horizon analyzes system performance, makes predictions and sends automated alerts but also advises and makes (automated) decisions for the farmer. “This allows our technicians to analyze system performance and identify machine issues. It also allows us to make predictions that help the farmer run his business,” explains Voogt.

Continuous innovation and experimentation are key principles of Lely’s data approach. Voogt emphasizes that this has been and still is a learning process. “While machine learning models can automate many tasks, farmers should still have the opportunity to make decisions themselves. On the other hand, just dumping a lot of data on the farmer isn’t helpful either. Striking the right balance between automation and control is essential and relies on individual farmer preferences. For simpler tasks and settings, though, automation makes much sense.”

Data analysis allows Lely to optimize the milking process for each cow on an individual basis. “Factors such as udder cleaning and preparation, stimulation duration and milking duration are fine-tuned to cater to each cow’s specific needs. We strive to have the best situation for each cow and her well-being. Some cows want to be milked three to four times a day, while others only want to be milked once,” explains Voogt.

Microservices

Lely’s software team operates under the Agile approach, deploying a new release of the Horizon software every two weeks. Hoogweg stresses the importance of seamless deployment to the thousands of farms using the software. “They rely on PCs that also run other software. Thus, maintaining high-quality software delivery without downtime is crucial.” The Horizon software on the PC is a .NET application with a web interface based on the Angular front-end framework.

Based on a previous application that was built monolithically, Horizon is now being transitioned into a modular system, explains Hoogweg. “My team is organized into domain teams that work on topics such as feeding, milking, health and reproduction. Our goal is to empower these teams to work autonomously, enabling them to test and deploy their own software components.”

While this transition is still a work in progress, Lely has successfully implemented microservices in the cloud portion of Horizon. This allows individual components to be upgraded separately using the container orchestration tool Kubernetes, while deployment pipelines have been automated with Gitlab CI. “On the farmer’s PCs, that’s a bit more difficult,” admits Hoogweg, “as we don’t have full control over such a PC. But we’re also looking at services to manage the farmer’s network and IT systems. Using those, we should be able to deploy various components locally as microservices in containers.”

Freedom to innovate

Lely’s workforce has evolved significantly over time, with an increasing number of electronics, software and data engineers joining mechanical engineers in the company. “Two-thirds of our product development department are software and data engineers now,” observes Voogt. “This shift highlights the growing importance of software in modern farming practices.” The product development team has also experienced substantial growth, expanding from 150 engineers five years ago to 500 today.

Lely introduces new innovations regularly to make the lives of farmers sustainable, profitable and more enjoyable, also for next generations. Voogt emphasizes that Lely has been actively involved in developing solutions to reduce nitrogen emissions long before it became a major concern. The company’s efforts include a system for separating manure and urine on stable floors, resulting in reduced nitrogen levels: on average, up to 77 percent ammonia reduction (certified by the “Regeling ammoniak en veehouderij” or RAV). Additionally, Lely is working on integrating emission measurements into Horizon, enabling farmers to provide evidence of their nitrogen reduction initiatives.

With 8 percent of its revenue invested in R&D, Lely fosters a culture that encourages employees to pursue their own ideas and innovatively tackle challenges. “If you have a good idea, the company gives you the opportunity to work on it,” asserts Voogt. “There’s a clear commitment to allowing sufficient time for research and development, even if it doesn’t yield an immediate return on investment. It’s a family business with a long-time horizon and social motivations. However, if you start something, you need to finish it. Lely motivates employees to take ownership of their projects. If you have an entrepreneurial spirit, you get a chance here to make a positive impact on people’s lives.”

This article was written in close collaboration with Lely.