Design and Development of an Automated Hyperspectral Imaging System for Plant Monitoring

Automated plant monitoring systems have emerged as a result of the growing demand for data-driven and sustainable farming methods, particularly in controlled settings like greenhouses. These systems are now crucial for improving our use of resources like water and for managing the problems caused by climate change, like the occurrence of droughts. The main goal of this thesis is to design and construct a prototype system that integrates hyperspectral imaging and automated movement. By tracking plant health and identifying early indicators of drought stress, this system can be used practically to study plant physiology and help develop more efficient farming methods.
The two primary components of the system are an integrated camera setup and a custom designed linear motion structure. The cameras are moved along a linear guide by the mechanical framework using an Arduino microcontroller and a stepper motor. Because of this, the system can automatically position itself over various plant pots, guaranteeing repeatable and consistent measurements over time. In addition to a stereo RGB camera to record the structure of the plants, the camera setup consists of two hyperspectral cameras that cover the visible to near-infrared and short-wave infrared ranges. Specialized software manages calibration and data collection for these hyperspectral cameras. The system can detect minute alterations in the plants that are connected to stress, pigment levels, or water content by recording comprehensive spectral data—long before any problems are apparent to the human eye.
At Fraunhofer IGD, the entire system was put together and tested in a greenhouse. To ensure that everything operated and that the cameras could consistently gather spectral data in a greenhouse setting, static tests were conducted and results validated. In the end, this work provides a versatile solution that connects advanced sensing, automated control, and mechanical design. It lays the groundwork for further enhancements, like incorporating machine learning to automatically analyze the data, adding machine Python scripts to automate the measurement process, boosting automation, and testing the system in various developing environments.