The world’s population is expected to increase by another two and a half billion to 10 billion by 2055. At the same time, the amount of arable land is expected to shrink as increased urbanisation takes hold. And that’s before other issues such as climate change are factored into the equation.
The end result is a lot of mouths to feed and no obvious way to provide the substantial increase in food production required.
However, one potential solution is vertical farming: something being explored by several startups. The list includes Dell-backed AeroFarms. The New Jersey enterprise combines IoT with data analytics to grow crops in a high-tech warehouse environment.
Arguably, more effort goes into harvesting the thousands of data points collected than the end result. Every possible metric is tweaked to produce the perfect crops and eke out every drop of potential from the resources used.
But these methods need to be perfected if there is any hope in scaling them and winning the public’s trust.
Which is where researchers from Microsoft, Intel, and Tencent, among others, come in. Several artificial intelligence teams are taking part in a vegetable-growing competition like no other. The Dutch university of Wageningen is currently hosting the Autonomous Greenhouse Challenge to explore artificial intelligence applications in indoor farming.
Growing cucumbers with AI
The contest, which launched in September and will culminate at the end of the year, is being funded by the Tencent Exploration Team. The division was set up in 2014 by the Chinese internet giant to invest in promising startups and work to solve problems that are global in scale.
The challenge facing entrants is simple: produce a cucumber crop within four months inside the Wageningen University greenhouses. The catch is everything has to be done remotely. The warehouse is completely automated.
The teams taking part, from the likes of Microsoft, Tencent and Intel, were allowed to set up their own sensors and cameras.
Models and machine learning algorithms are tracked, adapted and applied by each team to switch up the control settings for the next day or time period.
The winners will be selected in December based on a number of factors, ranging from crop health to resource efficiency and the robustness of their AI model.
Sustainability is a key part of the competition, so the entrants are looking for marginal gains and sophisticated techniques to improve outcomes, rather than simply maximising the resources (water, light and heat, for example) at their disposal.
So, why cucumbers? Tencent chief exploration officer David Wallerstein explained that they made the perfect test crop because of the amount of existing modelling data and research already available.
“This isn’t about snipping the cucumber off the vine and then putting it in the plastic bag or something … the entire greenhouse is like a robot, so it should be fully autonomous,” he told VentureBeat.
“The sensors are there, then the AI and intelligence is directing the actuators in real time how much nutrition to deliver. Do they play with lighting? Do they play with gases? And so on and so forth.”