Abstract
The continuous advancement of in planta sensing technologies is essential for sustainable agricultural practices in the era of Agriculture 4.0. Reliable, real-time monitoring of plant physiological status is critical for optimizing resource use, improving crop yields, and ensuring food security. This thesis addresses these challenges by using novel bioimpedance-based techniques for plant health assessment, integrating machine learning for stress detection, and utilizing agricultural waste for bioelectronic applications. These contributions align with the United Nations Sustainable Development Goals (SDGs), particularly SDG 2 (Zero Hunger), SDG 9 (Industry, Innovation, and Infrastructure), and SDG 12 (Responsible Consumption and Production). Among proximal sensing techniques, electrical bioimpedance spectroscopy (EIS) has emerged as a promising tool for monitoring plant physiological responses to environmental stress. However, its widespread application remains limited due to the challenge of continuous monitoring, which consequently restricts its studies to well-designed experiments for plant physiology under controlled environmental conditions. This limitation affects the quality of the data obtained. Additionally, the need for sustainable electrodes remains a significant challenge in the practical deployment of bioimpedance techniques. This thesis proposes a multi-faceted approach to overcome these limitations through the monitoring of iron stress and water stress of tomato plants in environmentally controlled conditions continuously using bioimpedance, the use of circuit parameters and machine learning for stress classification, and the fabrication of biodegradable bio-based conductive materials derived from tomato plant waste. Bioimpedance measurements were conducted in a controlled greenhouse environment, where over 8000 measurements were collected over a 38-day period under different stress conditions. The data were analyzed using impedance spectra to extract key circuit parameters relevant to plant physiological responses. Different supervised machine learning models, especially multilayer perceptron (MLP), demonstrated high classification accuracy, achieving F1 scores of 0.89 for water stress detection and 0.93 for iron stress detection. These findings highlight the potential of bioimpedance-based techniques for precision agriculture. In parallel, the research explored the development of biodegradable bio-based conductive composites out of the tomato waste for potential bioelectronic applications. Two classes of materials were fabricated: graphene nanoplatelet (GnP)-reinforced latex composites, which exhibited low electrical resistivity (0.46 Ω·m at 20% GnP concentration) and stable impedance properties, and graphene nanoplatelet (GnP)- pectin-hydrolyzed tomato waste composites, which demonstrated strong sensitivity to environmental factors, indicating potential use as humidity or temperature sensors. Moreover, preliminary tests showed that the latex-GnP electrodes provided reliable bioimpedance measurements over 170 hours. By integrating in planta bioimpedance sensing with sustainable material development, this research contributes to the advancement of smart agricultural technologies and green electronics. The findings support the transition toward data-driven, resource-efficient farming practices while promoting the utilization of agricultural waste for functional materials. This work lays the foundation for future innovations in plant monitoring, bioelectronic sensors, and circular economy-driven material science, aligning with global efforts to enhance food security, technological sustainability, and responsible resource management in agriculture.