In this column,Colinda de Beer, Senior Business Developer Horticulture at InnovationQuarter, writes about the road to hands-free production of greenhouse crops and the need for cooperation between tech developers and growers. According to De Beer, the majority of flower and vegetable growers in the Netherlands are forward-thinking companies that always want to stay ahead and need to keep innovating to maintain that lead.
Photo credit: Colinda de Beer / Innovation Origins
But why should a grower invest in these kinds of technological solutions and why can’t a tech specialist do it themselves? According to De Beer, a major reason for this is that many of the parties that have that technological knowledge do not know enough about horticulture. If you, as a robot builder, are thinking about developing a harvesting robot for tomatoes, for instance, the first question is: For what kind of tomatoes? Another important reason is a financial one. Companies that have the technical knowledge and access to horticulture have to deal with hefty development costs. I recently heard a developer say that development up to a working prototype costs at least 10 million euros. This is not an amount that the average tech developer can readily afford to pay. Collaboration is therefore essential for working on the right problem and for bearing the financial risk together.
GearRover – AI in the greenhouse
Product developers, as well as entrepreneurs in greenhouse horticulture, are still searching for a way to go about this. Is this actually working anywhere? Yes, fortunately, there are already examples where tech and growers are working very closely and iteratively together to figure out what is needed and what is possible. A number of entrepreneurs in the ornamental horticulture sector have teamed up with a tech company to develop a harvesting assistant. Previously, a logical question would have been “Can you make a robot that can cut roses as well as people can?”. However, due to the cultivation method and structure of a rose crop, this would have been a very complicated task that would have cost a lot of time and money. By examining the entire process, they jointly came to the conclusion that there would be a great deal to gain if a harvester could be helped to recognize the proper ripeness stage. The developed robot (or rather cobot in this case) locates the rose that needs to be harvested, then the employee actually snips it off and puts it on the trolley. The system uses vision technology and machine learning to do this.
You could be forgiven for thinking that the product is not yet finished, given that it does not harvest roses itself. However, the situation changes when you realize that it takes an employee up to three months to perfectly recognize the right stage for harvesting. That right stage has a lot of influence on the shelf life and consequently the quality of a rose. Once you know that, you then see that it makes perfect sense to first focus on a product that merely points towards the roses. Such a product already adds a lot of value and can already be sold then. Meanwhile, work is underway to add more functionality, such as a harvest prediction system and being able to recognize diseases based on images that are already collected during the robot’s work process.
To read the complete column, go to www.innovationoriginis.com.