About the Autonomous
Greenhouse Challenge

Goal of the Challenge

The goal of the challenge is to produce a cherry tomato crop within 6 months at a high level of production and a high resource use efficiency. For that, selected teams will get the possibility to operate a greenhouse compartment at the greenhouse facilities of WUR in Bleiswijk, The Netherlands. Teams need to achieve the goal by growing the crop remotely controlled, supported by measured values of greenhouse climate and crop development and also webcam footage.
The teams will have to make choices with respect to the control settings in order to control the crop production and quality growth remotely. They can also add their own sensors/camera’s to generate additional information. Each team will be able to extract necessary data from the greenhouse compartment and couple it to their own ICT/models/machine learning algorithms in order to decide on the control settings for the next day/period. They will send the control settings back to the system (the greenhouse climate computer) in order to steer the actuators automatically or send instructions for crop handling in order to reach the goal. WUR will continuously measure performance criteria per compartment and share them with each team and the public.
Show your skills, enjoy a multi-disciplinary collaboration and have fun in an international setting of experts!

Invitation, Rules and Regulations

Invitation to participate in the international challenge.pdf
Criteria rules and regulations.pdf

Results of the Challenge

Pre-challenge Hackathon


Final Event

Team name
Team Members
Team Composition (20%)
Strategy and AI approach for the growing challenge (30%)
Obtained results / reaching the goal in Hachathon (50%)
Total score


Total score 88.8
A team consisting of staff, researchers and students from Wageningen University, Evertill Co., Ltd, NXP Semiconductors, IGMPR Flower, Parks & More, Ibeo Automotive, Amsterdam UMC, CGI, Rotterdam.AI, Port Of Rotterdam-Fordata; a large part of the team also took part in the previous challenge
Technology is booming in this era. Agriculture cannot be left behind. We believe AI should be able to help human society in broader sense. With expertise of real tomato growing experience, plant science (greenhouse horticulture, plant physiology and crop modeling), Algorithm/Software Engineering, Applied Mathematics, Data science, Computer science and AI, our team hopes to improve the efficiency in greenhouse production and find a sustainable way to feed the world.
Our principle of decision-making is based on the real greenhouse growing experience, model simulation result and AI deep learning outcome. By using historical weather data for the period of real challenge, we will first simulate a reasonable crop strategy among the production period. Then we apply the climate and irrigation strategy automatically for the first two or three month according to the combination of weather forecast and expert experience. When after the first few trusses of tomato are harvested, the AI may receive enough data feedback from all source and start to generate outcome. The outcome of AI will be evaluated by human tomato expert and tested by tomato growth model before it validates to the real greenhouse control.
From the experience of previous challenge, we noticed the feedback of plant growth for once a week was too slow. Sensors that may reflect the plant growth dynamically will definitely help with the AI deep learning. We may apply several sensors that related to the plant fresh weight, leaf temperature and micro climate within different part of the plant.

AICU Team members and company.pdf (225,52 kb)


Total score 84.7
A team consisting of employees, researchers and students from A-net, Hankyong National University, Seoul National University, Samsung Electronics, University of Liege, EZFarm, FARM8, Spacewalk, ioCrops.
Artificial intelligence (AI) is making changes in many areas of our lives today, including the agricultural sectors. Even in agriculture, powered by AI-equipped cameras and sensors, once considered as an (almost) impossible tasks are being solved. Then what should agriculture look like in an age of AI? In order to find an answer to this question, we, Team Digilog (Digital+Analog), decided to participate in this challenge, Autonomous Greenhouses International Challenge (AGIC) 2019.
Teaming up with three universities and five companies, Digilog is united with a passion for writing a new history in the area of AI and agriculture. Our goal is, of course, to win the challenge, and thus we pursue to make technological advances in agriculture. However, at the same time, we also seek harmony between technology and human beings, of which many of us so far neglected to pay sufficient attention.
In the near future, we aim to host AGIC Asia or Argo-AI Challenge Asia in Seoul, Korea. Participation in this challenge will bring invaluable experience and vision to the running of AGIC Asia.

DIGILOG Team members and company.pdf (226 kb)


Total score 74.4
A team consisting of employees, researchers and students of CAAS, National Agricultural Science & Technology Center, National Engineering Research Center for Information Technology in Agriculture, Sichuan Film and Television University, Syngenta Seeds B.V.
We believe that autonomous greenhouses are the future of indoor farming systems. And we also believe that the artificial intelligence can bring actionable, data-driven insights to greenhouses and other indoor farming systems, for example:
  • Shorten the trouble shooting time of growers
  • Help them to become more strategic
  • Making foreseen decisions
  • Push the envelope of this tight-margin industry further
However, there are still many unclear questions to answer and a long way to go to achieve the final goal.

Bridge the gap

Effort is needed to bridge the gap between artificial intelligence and greenhouse horticulture, academics and industries. But the most important questions in this regard are “what“ and “how”.
We could have a long debate on topics like:
  • How AI can contribute to existing indoor farming systems? How reliable it is?
  • Why should we believe in a machine we build under the circumstances that man-kind does not have everything under control?
  • Or changing a perspective, can humans adapt the unpredictable environmental change as fast as AI?
We may have thousand questions and doubts, but at the end of the day we need to start somewhere. Eventually, we will find “what” and “how”.
IUA.CAAS is ready. And you?
Contact us

If you have any question, please do not hesitate to contact us.

Contact person: (team captain)

UA.CAAS Team members and company.pdf (273,01 kb)

The Automators

Total score 68.4
A team consisting of staff and students from Delphy, Wageningen University, 30 MHz; a large part of the team also participated in the previous challenge.
During last year’s autonomous greenhouse challenge, Delphy was part of the team The Croperators. One of the most important takeaways was that growing high quality and profitable crops autonomously is already possible. The jury stated that we developed a model that is directly applicable in practice. This gave us even more motivation to further develop this.
This year we're back as the Automators, combining crop-level knowledge with cloud platform engineering. Delphy, which has more than 200 consultants supporting growers and managing crops across the world, teamed up with 30MHz. Its platform ZENSIE ingests and analyses data interactively and enables users to continuously improve the production process of vegetables, plants, seeds, and bulbs.
By using data from the greenhouse and the crop, growers will be able to manage larger-scale cultivation. Artificial Intelligence will help the grower in taking strategic and operational decisions. The role of growers will change, mostly it will be monitoring the autonomous actions of the system and only in the most exceptional cases, they will have to interfere.
The worldwide challenges, which were the motivators last year, still exist. The world population is increasing rapidly. Besides that, we recognize the increasing lack of growth expertise in all parts of the world. We still believe that the solution to these challenges is a new way of managing greenhouse cultivations.

The Automators Team members and company.pdf (248,61 kb)


Total score 64.4
A team consisting of employees, researchers and students from Van der Hoeven Horticultural Projects, TU Delft, Keygene, Hoogendoorn Growth Management.
Van der Hoeven and Hoogendoorn build and automate high tech greenhouses around the word, feeding the growing human population with higher quality products. The number of such greenhouses increases much faster than the number of available skilled growers required to operate them. This bottleneck can be narrowed by further automating greenhouses. Along the way output quality and quantity, reliability, and resource efficiency can be improved. The companies are joining forces with TU Delft, which has an impressive track record on AI. TUD is currently organizing itself to better serve the horticulture industry with its expertise. Performing in the challenge is a way of further developing this technology to bring it to actual greenhouses as fast as possible.

Automatoes Team members and company.pdf (231,79 kb)