Artificial intelligence can help reveal the causes of stress in plants
As part of the European Digital Innovation Hub Ostrava (EDIH Ostrava) project, Signal Lab experts have developed a unique robotic system for camera-based plant inspections in indoor growing conditions. Thanks to early diagnosis, it will be possible to adjust growing conditions more quickly and increase crop resistance. They are also collaborating on the project with experts from the Institute of Agricultural, Food and Environmental Technology at Mendel University in Brno.
"Within the project we have successfully implemented a remote monitoring and sensing system for plants in indoor growing conditions, which was requested by SOHE. We developed software and implemented hardware for plant monitoring, including analysis of available solutions, design and build of a growing tent with controlled lighting, ventilation and irrigation. The solution also included the development of a climate monitoring system and the implementation of a camera system to enable remote monitoring. At the same time, we developed a mobile and desktop application for visualization of the data obtained. The application includes the proposed architecture, API, backup system and is fully integrated with the greenhouse control system. The result is a comprehensive solution that enables effective monitoring of plant growth and optimization of growing conditions," explains Radek Martinek from the Faculty of Electrical Engineering and Computer Science at VSB-TUO.
The research focuses on different types of vegetables grown in the laboratory conditions of the Faculty of Agronomy of Mendel University. Scientists are simulating conditions that can cause abiotic stress to plants. "In the growing tents, we can measure and subsequently regulate the internal environmental conditions relatively accurately. This allows us to, for example, briefly increase the temperature to 30 °C, which puts the plant under stress. Another feature that significantly expands the possibilities for research is hydroponic cultivation, where we can mix the nutrient solution as we need it. In this way, we can, for example, stress the plant with a lack of calcium, phosphorus or nitrogen," says research leader Vlastimil Slaný from the Institute of Agricultural, Food and Environmental Technology.
Scientists record the response of plants to stress conditions using various imaging technologies.
"We have developed a robotic arm with a hyperspectral camera that can measure various parameters, monitor plants at several levels and record even small changes caused by long-term stress," says Zdeněk Slanina from the Faculty of Electrical Engineering and Computer Science at VSB-TUO.
Once the researchers have collected enough image data, they will use it to train advanced neural networks.The artificial intelligence will compare the plant images with a large database and can determine the cause of the stress."Our goal is to create a digital twin of the plant, a virtual model that mimics its behaviour in real time.This will allow us to detect problems early and we can implement appropriate protective measures more quickly.
We plan to use drones and satellite imagery in the field," adds Zdeněk Slanina.
In the next phase of the research, the experts also want to focus on data acquisition in plants infested by pests and the development of a multisensor system for collecting not only image data.
PHOTO: Institute of Agricultural, Food and Environmental Technology AF MENDELU.