When a plant begins to experience stress, we humans usually don't see it right away. And as soon as we do see it, it may already be too late to save such a plant. This is especially true for freshly sprouted plants (cresses). They have a short cultivation period, so early stress detection is important to be able to correct. Gearbox, together with Koppert Cress and Division Q, is developing an innovative AI tool that should make it possible to detect plant stress with the goal of a better harvest.
The founders of Gearbox, Simone Keijzer and Johan Kreeft, are themselves from the Westland region and thus grew up among the agricultural and horticultural farms. Indeed, Keijzer is a horticulturist's daughter and thus saw from childhood how her parents did and had a lot of work done by hand. At a young age, she wondered if that couldn't be done differently. That feeling was intensified considerably after she met Cancer, who specialized in robotics. They were already life partners and decided to become business partners in addition. Their goal: to introduce digital technology-based automation into the horticultural sector. And that led to the founding of their company in 2016: Gearbox Innovations. Since then, the company has grown to over 40 people. So the market seems to be ready for this as well.
Inventing, designing and manufacturing smart technological solutions that enable garden and agricultural businesses to do their jobs more efficiently, better and faster. That's Gearbox's mission. In doing so, they use computer vision, artificial intelligence (AI) and robotics. One of the successes so far is the GearVision: a "digital inspector" that uses cameras to image harvested produce, such as cucumbers and vine tomatoes, and then uses smart software to run image analysis on them. In this way, the quality of the product can be quickly assessed. GearVision also indicates what can be improved during the growing and harvesting process. This often provides interesting insights that can lead to process optimization and labor savings.
The next step? Not just assessing the quality of a product afterwards, but checking whether conditions are optimal already during the growing process. "We are currently investigating whether it is possible to use cameras and an AI tool to detect whether or not a plant is experiencing stress," Harm van Adrichem explains. He started as CFO at Gearbox more than two years ago, actively engaged in business development. "And this project gained momentum when we came into contact with Koppert Cress: a producer and supplier of unique cresses that find their way into top restaurants and more exclusive vegetable stores worldwide. They are very interested in innovative methods to take the quality of their product to an even higher level. In this regard, it is also significant that they have founded a new company, Division Q, which focuses entirely on their sustainable innovations. So a collaboration with Koppert Cress and Division Q was obvious."
After that, things moved quickly. Early last year, Gearbox, Koppert Cress and Division Q applied for the MIT scheme. Van Adrichem: "With research projects, of course, you don't know in advance whether it will actually lead to a commercially interesting solution. So we were happy to take advantage of the MIT scheme supported by the NL AIC in 2022 to cover part of the development costs. That grant is really needed to make this kind of research possible. We were therefore delighted when we were told in the middle of last year that our application for the MIT scheme had been honored."
For now, the research is taking place in the laboratory. Koppert Cress has delivered several cresses to Gearbox to subject them to different types of stress. The assumption is that a smart algorithm can analyze the light reflected by the plants, making it possible to detect stress at an early stage. "Compare it to skin in the sun," says Van Adrichem. "When the skin turns red, you're already too late. So you have to make sure you seek shade before it gets there. And with plants, too, it's important to take action in case of stress before damage occurs. In the first phase, we focused on demonstrating stress in a laboratory setting. Now we are at the stage where we want to translate this to the greenhouse environment."
Should it turn out that the stress model also works in the greenhouse, it immediately opens up a world of possibilities. Cultivators can then be sent out with specific instructions. For example, that a plant needs extra water, or that a plant is suffering from pests and needs action.
"That would indeed be great, if all that becomes possible in the future," agrees Van Adrichem. "Especially also because with a targeted approach you can increase the overall quality and reduce the resource and energy consumption per cress. This helps Koppert Cress in their goal of one hundred percent sustainable business. We also want to eventually investigate whether it is possible to make this technique available for other crops, so that the positive impact on the environment can be further increased."
Visit theGearbox Innovations website. Or ask your questions to Harm van Adrichem.
SMEs, including many startups, have an important economic and innovation role. Research shows that SMEs face many barriers in applying AI. The AiNed MIT AI call to bridge the gap between the knowledge base and application of AI innovations is supported from the National Growth Fund Programme AiNed. This scheme will be repeated annually through 2026. Our website will keep you informed.