If there’s anything the pandemic has shown us, it is that more activities can be done from a distance than we’d previously thought. Meetings, family parties, shopping, and indeed, growing lettuce. Or at least, that’s what this year’s edition of the Autonomous Greenhouse Challenge is aiming to prove.
The autonomous Greenhouse Challenge is organized by Wageningen Research to show to a larger public the achievements of greenhouse automation. “Every year we add an extra step,” Silke Hemming, organizer of the event, explains. “We started the first edition with cucumbers. In the next edition, we went for tomatoes – a crop that can be controlled on quality demands. But we still saw that sometimes, a “human in the loop” was necessary during the competition. For instance, a human eye was needed to change certain climate settings during the process so the algorithm couldn’t take care of it all.”
The hackaton teams of a previous edition. Photo provided by WUR.
Real-time audience presence
This year’s challenge will feature lettuce. “Lettuce takes a shorter period to grow. Besides, it is easier to grow with no human interaction. We thought this was the perfect start for this year’s Autonomous Greenhouse Challenge,” Silke explains. “We expect teams to be able to grow lettuce with a fully automated algorithm in the greenhouse.”
The word autonomous here should be taken literally: after an initial “first test crop cycle”, the teams will be given the opportunity to update their algorithm for the second crop cycle. This second cycle will count for the challenge, and during this cycle, the algorithm is not to be changed anymore. And on top of that, the public will be able to follow the event via a website, we have e.g., 3D RealSense camera’s and Sigrow stomata camera’s to observe the crop. Next to that, all climate parameters and resource use will be published. In the end, teams will be judged on net profit. “As the algorithm cannot be changed during the contest, the public can follow real-time and teams won’t be able to benefit from each other’s data.”Photo provided by WUR.
The story behind automatization
Although robotization is a hot topic in horticulture, this project is not about automated labor. “We want to tell the story behind automatization, and how teams achieve this. Lettuce is highly automated already, and besides, lettuce is an important product globally, also in vertical farming. We want to educate a worldwide audience on the benefits of automated crop growth control.”
One of the benefits of the challenge is that all data sets created by the teams will be released to the public, as we also did in former editions. “I know several companies that use this data set to develop their own algorithms,” Silke adds. “In general, we see there’s a great need for data and tools to develop optimal growth strategies to increase crop yields. There is also a shortage of growers who possess the expertise to run a large greenhouse company. With an autonomous system, one person can run multiple greenhouses remotely.”
At the same time, lettuce also poses challenges to the teams. “Available greenhouse space is a major constraint they have,“ says Silke. “Optimizing the use of space is paramount because lettuce grows from a little seedling to a huge head of lettuce. And during this challenge, we don’t work with vertical solutions. So, the teams’ algorithms also have to decide on spacing the crop to use all the available light, space, and therefore resources.”
Photo provided by WUR.
Machine learning
The first part of this year’s edition takes place from 1 June to 14 July as an open Online Challenge, aiming at testing machine learning and computer vision skills of participants of the AI community. The winning team will receive a wild card for the next rounds immediately.
The previous editions of the challenge already had some far-reaching consequences. “During the first challenge, one of the teams joined and founded Blue Radix, a company that offers algorithm-based solutions for greenhouse growers. Other participants came from Delphy. They have now developed their algorithms further for their consultancy company. In other words, teams keep the knowledge they established during the contest, and apply that to improve their business.”
Photo provided by WUR.
For more information:
Wageningen University & Research
www.wur.nl