Automato harvest: AI beats farmer in greenhouse keeping

Source: Xinhua| Published: 2020-06-18

Three greenhouses in northeast China's Liaoning Province had a bumper crop of tomatoes in late May.

Local tomato growers paid tribute to their new helper: artificial intelligence (AI).

The iGrow AI system produced higher yields and profits and lower labor costs in two greenhouses than the one run by experienced farmers.

Farming is highly complicated and crops have different requirements in greenhouses during their different growth cycles, said Liu Jianhua, a senior agricultural technician and consultant on the pilot greenhouses.

Temperatures for tomatoes need to be different during seeding, flowering and fruiting.

Farmers usually estimate a fixed temperature based on their experience, so it's hard to achieve precise, real-time and automatic control of the greenhouse and the optimization of planting, said Liu.

However, the AI system uses cameras and sensors to collect data every minute or hour on the temperature and humidity of air and soil, the concentration of carbon dioxide, photosynthetic effects and other factors.

According to the Chinese tech giant Tencent, which developed the iGrow system, the greenhouse simulator can simulate 82 growth cycles in 15 seconds. Planting decisions are made using its AI algorithm, which is continuously fed new data and tweaked as needed.

The system automatically controls the greenhouse environment, precisely targets resources and improves quality and yields of tomatoes while farmers need to do only the basic work of planting, picking and daily maintenance.

In 2018, the developers of iGrow took part in the Autonomous Greenhouse Challenge. Computer scientists and horticulturists in China used the system to remotely grow cucumbers in a greenhouse in the Netherlands.

The iGrow system ranked first in AI decision-making and second overall in the contest.

Two years later, the system learned how to grow tomatoes.

In February, the three greenhouses were hit by a sudden cold spell. In the control greenhouse, farmers set up three different fixed temperatures as usual, but fluctuating conditions affected the growth.

Meanwhile, the stability and disaster tolerance of the AI algorithm was well demonstrated in the iGrow greenhouses. The temperature and other indicators were adjusted every hour using automatic rolling curtains and ventilation.

Tencent said the pilot project in Liaoning Province verified the feasibility of AI planting in China, and it was looking to further develop smart farming. 

LIKE|0
中国好故事
0:00
0:00