Keeping tomatoes healthy

Rennigen/Tokio: Farmers lose their harvests because they don’t know the right timing of spraying. Now, growers can benefit from a Bosch project in Japan: Plantect is a smart solution that analyzes sensor data using artificial intelligence technologies to optimize plant growing in greenhouses. The project is part of the Bosch growth initiative called Future with Japanese Innovation (FUJI), in collaboration with the Bosch Center for Artificial Intelligence (BCAI). “At present, farmers act based on their past experience,” project manager Ryosuke Suzuki says. “They use more pesticides than necessary.” Plantect uses sensors to measure the humidity, temperature, carbon-dioxide levels, and sunshine. Such environmental parameters are of paramount importance for plant growth. They are transmitted via a gateway (hardware) to a cloud server, where a system equipped with algorithms analyzes the data together with other important agronomic parameters as well as weather forecast and visualizes the results on a user-friendly app.

Nipping disease in the bud

The disease risk analysis solution, jointly developed by the FUJI and BCAI, leverages artificial intelligence to predict the need for pesticides. Ryosuke Suzuki: “Farmers can then see whether they really need the spray pistol.” It pays to check the app: With the current field test, Plantect shows a 66 percent reduction in the number of diseases as well as a 29 percent reduction in the usage of chemicals. With an accuracy of 92 percent, the system can detect whether there is any risk of infection in the air. Through the systematic use of chemicals, farmers can prevent the breakout of diseases such as gray mold. “It’s the same for plants as for people: infections are invisible. Once they are sick, all you can do is limit the damage. You have to nip the infection in the bud.”

Global cooperation

The standard version of Plantect, which provides real-time monitoring of environmental parameters, is suitable for all types of plants. The version featuring early detection of infections is only available for tomatoes at present. However, developers are in the process of adjusting the algorithm for additional species. Versions for cucumber and strawberry plants should follow by the end of 2018. Farmers appreciate the price model: the entry-level product is free of charge; only a monthly subscription fee is charged. Plantect uses wireless hardware and – depending on the size of the greenhouse – is made up of one or several sensors along with a central gateway that transmits data to the cloud for analysis. The battery-operated sensors can be installed anywhere in the greenhouse without any need for sockets, cables or other constructions.

If, for instance, a farmer wants to gauge whether environmental parameters differ between different sections of a greenhouse, multiple sensors can be installed accordingly. In their development work, Ryosuke Suzuki and his colleagues collaborated closely with the BCAI. The BCAI supported colleagues in Japan from the data compilation phase to implementation: “We advised the team in Tokyo, and we jointly developed the artificial intelligence algorithm especially together with FUJI agronomists,” says Vusirikala Nataraju, who is responsible for the project at BCAI. In 2018, Plantect will be launched on the Chinese and South Korean markets. Ryosuke Suzuki has confidence in his team’s work and the collaboration with BCAI: “We’re not just talking about it, we are putting our plans into action. That is allowing us to create products that will make people’s daily lives better.”