Artificial Intelligence‐Driven Robotic Sensing System for Noninvasive Crop Health Monitoring and Autonomous Irrigation Management

Abstract

This study introduces an artificial intelligence (AI)‐driven robotic system utilizing a 3D‐printed electrophysiological (EP) sensor for noninvasive, real‐time monitoring of plant health signals across different irrigation levels, highlighting the crucial role of these technologies in enhancing smart agriculture and sustainability. The sensing system consists of a mobile robot with a 3D EP sensor and portable Faraday cage for data acquisition, using an AI‐powered convolution neural network to analyze EP data in greenhouses and categorize irrigation levels to optimize water usage for scalable agricultural management. The findings reveal that the 3D EP sensor displays lower and more stable contact resistance (2.10 ± 0.52 MΩ) compared to flat thin‐film sensors (2.96 ± 1.45 MΩ), ensuring high electrical reliability due to effective contact with hairy tomato leaves. The 3D EP sensor's high sensitivity (signal resolution of 0.0122 mV) detects subtle EP signal changes linked to irrigation levels, aiding water optimization and crop yield enhancement. For the first time, this study employs scalogram images for detailed analysis of plant EP signals, achieving a classification accuracy of 86.91%, comparable to red, gren, and blue image‐based methods (86.37%). This system is a reliable tool for long‐term monitoring in smart farming and provides insights into plant signal dynamics.

Code and data availability

The supplied blocks describe the paper's EP sensor, greenhouse data collection, scalogram/CNN and EfficientNet analyses, but contain no data availability, code deposit, or public repository statements. No paper-specific public phenotype datasets, images, code, or trained models are identified.

No evidence-backed public reproduction asset is currently recorded.