Our Research

Modern agriculture heavily relies on sensors for yield improvement and monitoring. Sensors allow the detection, recognition, and classification of agricultural parameters and failures. Examples of beneficial sensing modalities are imaging, which showed its power in providing structural information, allowing disease detection by symptoms; spectroscopy, which was shown to help estimate changes in plants and water chemical compounds; and thermal imaging, which allows remote identification of water deficits. Indeed, agriculture could earn outstanding benefits from using sensors in the field. However, use of sensors in the agricultural field is challenging:
1. The nature of the objects in the agricultural scenery is variable and unstable, though they are hard to bind.
2. The sensing is done under changing environmental conditions, which affect both measuring conditions, plan behavior, and sensor response.
3. Economics limits the price for affordable electro-optics sensors for agricultural use, which results in a deficit in the robustness and stability of the sensing system.
To tackle that, we explore optical-algorithmic research, where knowledge of the physics of the sensing phenomena and the sensor's design serve as prior knowledge for solving machine-learning/inverse problem algorithms that support the acquisition problem and provide the required information in this challenge conditions.