Plant Growth Monitoring, Prediction, and Self-Regulation Utilizing MXene/CNTs/TPU Flexible Strain Sensors Integrated with Deep Learning Algorithms and Soft Actuators

• • Tensile gauge factors of 5.41, 7.39, and 3.39 across 0–20%, 20–50%, and 50–70% strain ranges, respectively, enable high-fidelity detection of plant stem elongation; the 7.39 peak in the 20–50% range matches the rapid growth phase of many crops, allowing early prediction of yield-limiting stress events with actionable temporal resolution. • • Bending sensitivities of 1.79, 0.89, and 0.46 over 0–30°, 30–90°, and 90–120° provide sufficient angular resolution to track leaf movement and turgor-driven shape changes; the 1.79 value at low angles is critical for detecting incipient water stress before irreversible wilting, reducing irrigation waste by enabling threshold-based actuation. • • Integration of an LSTM deep learning model with the tensile sensor achieves phased growth prediction, converting raw strain time-series into forecasted growth trajectories; this addresses the temporal resolution gap of UAV and infrared spectroscopy methods, which cannot capture individual plant dynamics at hourly or daily scales. • • The SMA-based soft actuator, driven by bending sensor feedback, forms a closed-loop plant sensing-actuating system that physically assists leaf growth; this self-regulation capability is absent in passive monitoring technologies, offering a pathway to mitigate environmental stress in alpine, desert, or space agriculture where human intervention is impractical.