The technology enabling the agriculture industry’s future

Bucolic scenes contain AI-powered robots, gently picking ripe fruit..
30 November 2023

Christian Baker, site manager / grower at Zordi, in front of the “automated irrigation system,” which disperses nutrients among the produce in the greenhouses at the farm. Photo by Lakota Gambill for The Verge.

Getting your Trinity Audio player ready...
  • AI agriculture is ready roll out in the food industry. 
  • Short labor supplies mean focus is on robots to pick up the slack. 
  • AI and machine learning help harvest and and fertilize crops all over the world. 

Besides its potential to take over the world and, in time, become our robot overlords and wage war on humankind (etc., etc.) AI has enormous capacity to benefit the agricultural sector.

In Japan, smart agriculture is becoming the norm. Producers are expected to trust artificial intelligence with more labor-intensive tasks to alleviate manpower shortages.

Large-scale greenhouse-based market gardening farmers have begun using AI-equipped robots in ways that seem to change the future of cultivating and harvesting agricultural so. Just the whole world shifting, then, at least, in Japan,

In September, within the plastic walls of a greenhouse on a farm in Hanyu, a four-wheeled robot rolled between lush leaves gathering only the ripest cucumbers. Takeshi Yoshida, head of the Takamiya No Aisai farm, spoke of initial fears the robot would cut off the cucumber stems, “but it moves accurately. We expect much from the robot now that labor is in such short supply,” he said.

Agrist’s AI robot is cool as a cucumber (sorry).

A subsidiary of the Takamiya farm, which manages greenhouses and other facilities, operates commercially, while Takamiya No Aisai is the first farm to lease one of theagricultural robots.

The robot was developed by startup Agrist and uses a camera and AI to determine whether crops are ready for harvest. Agrist has been developing harvesting robots since it was founded in 2019 in the Myazaki prefecture.

Using a camera mounted on the robot, it checks the size of cucumbers and recognizes ripe ones, harvesting between one and three vegetables every two minutes, putting them in a container. The robot can accurately position its arms so as not to damage cucumber stems. The hope is that the robot’s success will see more farms adopting systems of this kind.

In Kanagawa prefecture, an agricultural venture company in Kamakura called Inaho, has leased an AI-equipped robot to a farm in the Netherlands. It automatically picks cherry tomatoes in bunches or individually, depending on the mechanism used.

The AI analyses images and selects tomatoes that are ripe and easy to pick. The robot then uses its arm to harvest them. Developing a robot that can carry out this entire process is complex: tomatoes tend to bunch around the plants’ leaves and stems, necessitating a complicated mechanism to pick them without damaging the plant.

Inaho’s robot reaps about 40% of the available ripe tomatoes overnight, leaving the rest for harvest by humans during the day; presumably people who were once worried robots would steal their jobs!

The Netherlands, an agri-food powerhouse, is just the start for Inaho which hopes to export its agricultural technology around the world. The company has also developed a robot to harvest asparagus and plans to begin leasing the machines from April 2025.

The robot in action, picking toms. Courtesy of Inaho Inc.

“Although more time is needed to let robots harvest all crops, those currently available can sufficiently support farms with labor shortages,” said Soya Oyama, chief operating officer at Inaho.

According to Takanori Fukao, professor of robotics at the University of Tokyo, Japan is looking at a new frontier. “Starting with greenhouse cultivation where harvesting robots can move easily, examples of their introduction to open-field cultivation are likely to increase,” he said. “In the future, in order to make full use of robots, it’s likely that farms will have to be prepared, [..] taking into account the placement of crops in advance, for instance,” Fukao said.

On the other side of the world, Canadian potato growers are now using AI to monitor and predict the nutritional needs of their crops in real-time.

“As the Canadian potato landscape continues to evolve, the delicate equilibrium between production ambitions and environmental protection remains at the forefront of industry considerations,” researchers say.

In research trials, a portable spectrophotometer (optical sensor, to the layperson) is used to quickly determine petiole nutrient values in potato fields. Combined with machine learning algorithms trained on historical data, this technology enables almost real-time assessment of the individual plant’s nutritional needs.

For those of you not on top of your potato farming techniques, accurate understanding of the crop’s needs means fertilizers and other additives can be applied more efficiently and in a targeted manner, rather than with the usual gung-ho freedom that potato farmers once adopted.

Although there are dangers from AI in many arenas, the hype cycle of the mainstream media is definitely exaggerating them. When it comes to agriculture, though, AI can be – and is being – a force for good, providing nifty ways of bettering agricultural processes and taking the strain off farms suffering personnel shortages.

It might also facilitate sustainable farming.

In 2020, robotics expert Gilwoo Lee and a sixth-generation farmer, Casey Call, founded Zordi. Like Agrist and Inaho, Zordi is an agricultural platform blending robotics and AI with greenhouse growing. The startup is backed by Khosla Ventures and leverages traditional farming expertise and new technology to grow strawberries in the Northeast of America. From planting to harvest, robots do everything in the greenhouses – with human supervision, of course.

AI and machine learning monitor the growing process of unique varieties of strawberries imported from Japan and Korea and control the environment inside the greenhouses. Robots also harvest the ripened fruit.

According to Lee, success with the delicate fruit (strawberries need very particular growing environments) will mean the technology can be extended to other crops. “I think controlled environment agriculture or greenhouses, for us, is a very good way to feed the world with sustainably grown local fresh produce, and that was the mission that I wanted to see happen,” said Lee.

When AI and climate are together in a sentence, they usually sandwich “is really bad for,” but for the agriculture industry, machine learning means cutting down on technologies. Because of a need to understand vast amounts of information, from microclimate to soil pH, there’s a massive glut of information, and it isn’t standardized across the industry.

According to Vonnie Estes, vice president of innovation at the International Fresh Produce Association, until the advent of AI there was no way of parsing and using all the data necessary to improve yields and be more sustainable at any stage of a crop’s lifecycle.

“I think that from a climate-smart perspective, we’re just going to get more tools that are going to help farmers make better decisions to only use things — like water, pesticides, and other applicants — when they need to use them.”

AI alone isn’t a solution, but it is an enabler. Climate change, growing population and personnel shortages are all things to which farmers need to adapt; smart agriculture, powered by AI might just be the answer.