What does it take to make robots a success in agrifood?

Last week I attended the event RoboBusiness in The Hague. A specific part of the program was dedicated to the agrifood sector, and deservedly in my opinion. The various presentations confirmed to me the big potential that robots have in our industry. With an increasing demand for high-quality food, combined with resource scarcity, robots can be a key driver for solving the world’s food challenges. They can increase efficiency, improve yields and at the same time be the answer to various labour issues. Again I wondered why, then, it is taking so long for robots to really get big in agrifood.

by Mike Poodt

Even during my studies in Wageningen, some 20 years ago, we were already looking into using harvesting robots in horticulture. Yes, we have come a long way since then, but they’ve still not been adopted. And it’s not only in agrifood that we see a relatively slow adaption of technical possibilities; during the congress a video was shown of a housekeeper robot. The video was made some decades ago, and at that time they surely expected that by 2017 the robot would have become a reality. But just think about how many hours we still spend on domestic chores...

Photo: Rijk Zwaan

Challenges and solutions
Let’s first get into the specific challenges for agrifood. What makes the implementation of robotics so difficult for this industry? Obviously, this is mainly due to the natural character of the products:

We have to deal with natural variations in shape, which makes it hard for grippers and image analytics.
We have a huge amount of different growing systems and product combinations, so we need robots that are flexible and at the same time specialized in very delicate tasks.
Crops must be treated or harvested exactly at the right moment, while robots need high capacity to succeed.

However, we have seen that these challenges can be overcome. There are already some good examples of robotics in agrifood:

Lely has developed a milking robot. First introduced in the early ’90s, this innovation is now widely accepted and implemented.
We have mechanical weeders which can detect weeds between the crops and get rid of them. Robotic weeders can reduce usage of chemicals by up to 95%.
Precision agriculture: with the help of drones and GPS, robots and sensors enable us to achieve a whole new level of crop management. Thanks to data and sensor technology, we can now give each individual plant its most optimal treatment, resulting in more yield with less input.

It is hard to explain why these particular robots have become a success. Partly, it may be due to how urgent the need for efficiency was. But sometimes the key driver can come from a different angle. Take the robotic weeder: it is mainly thanks to demands from organic farmers that it has been developed and made a success.

Massive amounts of data
Apart from how they have become successful, the robots mentioned above have something else in common. They all produces massive amounts of data – not only data for their primary usage, but data that can be used for all kinds of other purposes as well. I think the challenge is to merge and analyse all this data to make actionable information out of it.

To illustrate this, let’s take the robotic weeder again. It detects not only weeds using image sensors, but also the crop. It can measure the size of the crop, the chlorophyll activity, the need for nutrition. It can detect diseases, and predict yield and the best moment for harvest.

If you think all these insights and data are only valuable for the farmer, you’re wrong. Retailers and the processing industry can greatly benefit too: they will be able to optimize their processes, based on better information about how much product of which quality can be expected in which week. From logistical processes to marketing campaigns, all processes can be optimized. Waste can be reduced and tracking and tracing can be lifted to a higher level.

At the very beginning of the supply chain, at breeding companies like Rijk Zwaan, this information can be very useful too. It’s a breeder’s dream to have so much performance data about their varieties. Traits and varieties can be further optimized. Based on genetics, they can even create varieties with robot-enabling traits. Think about leaf position, internodium length or peduncle length for easier image recognition. The huge synergy potential between breeding and robotics was clearly explained by my colleague Björn D’hoop in his presentation during RoboBusiness.

Another speaker was Richard van der Linde of Lacquey. They have invented a robot for the processing industry that can cut out the core of iceberg lettuce based on image analytics. It does this more efficiently than a human can, resulting in less waste. But what is the side effect? Every single crop is measured. Quality data is produced per lettuce head. Obviously, not only the processor can benefit from this data. Combined with weather and geographical information, it will enable the industry to jointly raise the quality and yield of a whole product category.

Intensify collaboration
We are still talking about relatively simple robots. But imagine what can happen if we combine the data from all robots throughout the supply chain. I am convinced that this will be the most powerful method to optimize the safety, quality and quantity of fresh produce and to keep feeding the ever-increasing world population.

Coming back to the main question of this blog: among the speakers at the congress, there was already a wide consensus about the importance of collaboration. Gert Kootstra from Wageningen University mentioned the triangle of robotics, agronomists and end users. His colleague professor Eldert van Henten emphasized the role of small, flexible companies to bridge the gap between robotics and agrifood. When Vincent Kwaks, CTO of Vanderlinde, was asked why he thought that robotics in agrifood is developing so slowly, he answered that ”initiatives seem to be too fragmented”.

I would add one more important aspect of collaboration: we need to bring the entire food chain together. As mentioned above, automatic data collection may be the key success factor for robotics in agrifood and this data is relevant for all chain players. We have learned that it takes multiple decades to successfully develop and implement robots in agriculture, so we should set our individual interests to one side and intensify collaboration – between universities, the robotics industry and all the members of the food supply chain.

And you know what… For each link of the chain, we have the absolute top of the world right here in The Netherlands. The best technical and agrifood universities, the world’s best breeders, the best growers, the best infrastructure. You name it and it’s in place. By utilizing this unique position, we can make a valuable contribution to the world food supply and at the same time make The Netherlands the world leader in innovative agrifood robotics. Will you join me in this fascinating challenge?

Mike Poodt is Business Consultant Innovations for international breeding company Rijk Zwaan. Spending 30% of its annual turnover on R&D, this Dutch family-owned company sells its vegetable seeds in more than 100 countries. One of Mike’s responsibilities is to work with and to design new technologies that can improve the primary processes within Rijk Zwaan. He studied Agricultural Engineering in Wageningen and has been working for Rijk Zwaan since 2003.