Israeli startup develops first AI robot for picking tomatoes

Growing up in a kibbutz and working in agriculture from a young age, Adi Nir, founder of Israeli startup MetoMotion, left the fields, as many others did, to make a living in the tech industry. The widespread global shortage of fruit and vegetable pickers is what brought him back to his roots to develop the world’s first robot for picking tomatoes.

Fewer and fewer people work in agriculture, which employs just 5% to 10% of the workforce in the European Union and euro area, and 6% in OECD countries, according to World Bank data. In Israel, only 1% of all Israeli workers are employed in agriculture, the data shows.

Israeli farmers are also struggling with labor costs. Since few Israelis work in agriculture, growers need to bring foreign workers to Israel to do the tough manual work, but are limited by how many permits the government allocates, driving up salaries.

“We hear many times about farmers leaving the crops to rot because there is no one to pick them,” Nir told The Times of Israel. “Today can’t grow tomatoes like 30 years ago — for them to be high quality and competitive in pricing, you need to do some transformation.”

After graduating as an engineer from Haifa’s Technion – Israel Institute of Technology, Nir worked for 16 years in the aerospace and defense industry, managing R&D operations and developing cutting-edge system technology.

Realizing that the shortage of manpower in the farming workforce and the related costs were unsustainable, and drawing from his tech experience in the aerospace industry, Nir founded the Yokne’am-based startup MetoMotion in 2017 to develop a robot that can do labor-intensive fruit-picking quickly and efficiently.

Since then, MetoMotion has raised a total of $10 million from investors led by Israeli venture capital fund the Trendlines Group, Netherland-based greenhouse technology company Ridder, and another Dutch investor.

“The idea was to help farmers solve the labor shortage problem as more and more young people are seeking more advanced occupations where they don’t need to use their hands or back,” Nir said.

“We used the capabilities of artificial intelligence to develop a platform which is designed to work in a greenhouse environment, where you can produce high quality yield in different weather conditions with much less water than you do outside all year around.”

“Unlike other robotic or industrial applications, which are characterized by more repetitive tasks, in agriculture every plant is a little bit different. So we came up with a solution based on AI capabilities to look at the plants, understand the structure, environment, how to measure the ripeness, and how to decide if it’s ready,” Nir said.

The startup’s driverless robot has two robotic arms to pick and harvest tomatoes on both sides of a row simultaneously in high-tech greenhouses. The autonomous guided vehicle is equipped with AI-powered 3G sensor vision technology that generates a map of the plant and its crop. It drives down the greenhouse rows with the help of sensors that can detect obstacles to avoid accidents.

The advanced vision system detects tomatoes ripe for picking and guides the robotic arm to the location, cuts and catches tomato branches in a single operation, and places the fruit on a conveyer belt before dropping it into onboard fruit container units, working at a speed of 16 seconds per cluster. The robotic arm clears away obstacles or hidden stems without causing any damage to the branches crop or plant. Once the driverless robot reaches the end of the row it stops and returns to the start of the row for its trailer to be removed and sent to the warehouse.

MetoMotion decided to focus on tomato harvesting as about 50% of production costs are labor expenses.

“More than 30% of the crops grown in greenhouses are tomatoes because we need them for a lot of things, for our salad, pizza, and ketchup,” said Nir. “Our robots are able to reduce the labor needs of harvesting by about 90% and cut production costs by about 50% by saving on manpower costs.”

A fleet of five robots working in a 50,000-square-meter greenhouse will need about a week for the tomato harvest, Nir said.

As the robots are cruising around the greenhouse for harvesting they also automatically collate data from sensors on the ripeness, the quantity, and weight and size of each tomato for better crop management. Using the data it is possible to predict the volume of output in each greenhouse to help generate yield forecasts and distribution analyses.

“All this data is uploaded on a dashboard in a cloud system,” Nir noted. “You can digitize many of the tasks that we are doing today like looking at the diseases, pesticide and things like that.”

For now, the tomato harvest robot is used at two greenhouse growers in the Netherlands. In Israel, the robot will be showcased for the first time this month at an agriculture conference. The startup, which employs about 15 engineers and agricultural experts in Yokne’am and four in the Netherlands, is currently seeking to raise $8 million in funds from investors to scale up production this year.

Nir hopes to install about 10 to 20 of the robotic machines this year.

“Our first target is in Western Europe — that is the Netherlands, France, Germany, Belgium, and all the countries around that region, and next is Canada and the US,” said Nir. “Apart from tomatoes, we also want to make use of the technology for different vegetables like cucumbers, eggplants, and peppers, which will also be available soon, because they are growing in the same environment.”

Although MetoMotion’s focus is currently on fruit and vegetables, Nir already has his eyes set on another industry.

“We see a lot of need coming from the cannabis industry, so this will probably be also added to our roadmap, but will require more modifications,” Nir said.

“Every week I get an email from a farmer that has another crop that needs picking, whether avocados, oranges, grapes, coffee or tea.”