A Generative AI-Assisted Piezo-MEMS Ultrasound Device for Plant Dehydration Monitoring

Area of Science:

• Agricultural Engineering
• Materials Science
• Biomedical Engineering

Background:

• Accurate plant hydration monitoring is crucial for agricultural productivity.
• Existing leaf water content sensors are often invasive, bulky, and power-inefficient.
• Micro-electromechanical systems (MEMS) offer potential for miniaturized, low-power sensors.

Purpose of the Study:

• To develop a CMOS-compatible, plant-leaf attachable piezo-MEMS ultrasound device (PLP) for real-time dynamic moisture monitoring.
• To enable rapid, non-invasive measurement of relative water content (RWC) in plants.
• To create a reusable sensor system that reduces electronic waste.

Main Methods:

• Fabrication of piezoelectric micromachined ultrasound transducers (PMUTs) using piezoelectric over silicon-on-nothing (PSON).
• Development of a CMOS-compatible, reattachable PLP device for non-invasive hydration monitoring.
• Application of generative deep learning (conditional variational autoencoder - CVAE) for signal translation to RWC.

Main Results:

• The PLP device demonstrated non-invasive hydration monitoring across diverse cultivars within a 70% RWC detection range.
• The generative AI-assisted system achieved a root-mean-square error of 1.25% for RWC measurement.
• The PLP device is reattachable, enhancing reusability and reducing electronic waste.

Conclusions:

• The developed PLP system offers a significant advancement in precision plant health management.
• This AI-assisted sensor enables direct correlation of plant responses to environmental changes.
• The technology promises improved irrigation practices, enhanced agricultural efficiency, and environmental conservation.