Mable P. Fok

Non-destructive methods for real-time phenotyping of root growth and development are essential to accelerate the breeding of climate-resilient crops. We present DIRTNet (Digital Investigations of Root Traits Net), a hybrid deep learning model that interprets in-soil signals collected by Fiber Bragg Grating (FBG) sensors. We first evaluated DIRTNet in a controlled simulation study in which metal rods of varying diameters were inserted into soil to emulate root growth. Using augmented training data with an 80:20 train–test split, DIRTNet surpassed baseline models, achieving 0.97 accuracy for root depth prediction and 0.94 for root diameter prediction. We then applied the best-performing models to maize (Zea mays) roots monitored by six FBG sensors placed at different soil location and depths. A control trial and a drought-stress trial yielded 192 samples, split 80:20 into training and holdout sets; the hold-out portion was divided equally into validation and test sets. Augmentation was applied only to the training data. On real maize root data, DIRTNet achieved 0.92 accuracy for both root depth and root width prediction across variable growth periods, covering root mass lengths up to 20 cm and diameters up to 10 cm. The prediction accuracy is independent of the soil moisture and drought condition. In a fixed-interval analysis, accuracy reached 0.97 for days 25–42 after planting and 0.93 for days 25–48, demonstrating robust performance as root complexity increased. DIRTNet also classified control versus drought-stress conditions with 0.87 accuracy. Together, these results establish DIRTNet as a continuous, scalable, and non-destructive approach for root phenotyping with strong potential to support crop improvement under changing environmental conditions.

Real-time measurements of crop root architecture can overcome limitations faced by plant breeders when developing climate-resilient plants. Due to current measurement methods failing to continuously monitor root growth in a non-destructive and scalable fashion, we propose a first in-soil sensing system based on fiber Bragg gratings (FBG). The sensing system logs three-dimensional strain generated by a growing pseudo-root. Two ResNet models confirm the utility of in-soil FBG sensors by predicting pseudo-root width and depth with accuracies of 92% and 93%, respectively. To analyze model robustness, a preliminary experiment was performed where FBGs logged strain generated from a corn plant’s roots for 30 days. The models were then retrained on new data where they achieved accuracies of 98% and 96%, respectively. Our presented prototype has potential prospects to go beyond measuring root parameters and sense its surrounding soil environment.

An inchworm-inspired soft robotic climber with embedded Fiber Bragg Grating (FBG) sensors has been designed and experimentally studied. Real-time motion reconstruction of the climber is achieved using FBG and Long Short-Term Memory (LSTM) neural networks.

Soil salinity content is monitored by embedding a distributed fiber optic sensor underneath soil and actively sensing an acoustic signal’s propagation properties. The system can distinguish soil salinity levels from 0 dS/m to 8 dS/m.

Real-time measurements of crop root architecture can overcome limitations faced by plant breeders when developing climate-resilient plants. Due to current measurement methods failing to continuously monitor root growth in a non-destructive and scalable fashion, we propose a first in-soil sensing system based on fiber Bragg gratings (FBG). The sensing system logs three-dimensional strain generated by a growing pseudo-root. Two ResNet models confirm the utility of in-soil FBG sensors by predicting pseudo-root width and depth with accuracies of 92% and 93%, respectively. To analyze model robustness, a preliminary experiment was performed where FBGs logged strain generated from a corn plant’s roots for 30 days. The models were then retrained on new data where they achieved accuracies of 98% and 96%, respectively. Our presented prototype has potential prospects to go beyond measuring root parameters and sense its surrounding soil environment.

Photonic signal processing is a broad and multi-disciplinary field, bringing together topics from physics, materials, and devices; communications, signal processing, and information theory; neuroscience; quantum information science; linear and nonlinear science; computing; and ultrafast science.

In this Letter, an unsupervised-learning platform—generative adversarial network (GAN)—is proposed for experimental data augmentation in a deep-learning assisted photonic-based instantaneous microwave frequency measurement (IFM) system. Only 75 sets of experimental data are required and the GAN can augment the small amount of data into 5000 sets of data for training the deep learning model. Furthermore, frequency measurement error of the estimated frequency has improved by an order of magnitude from 50 MHz to 5 MHz. The proposed use of GAN effectively reduces the amount of experimental data needed by 98.75% and reduces measurement error by 10 times.

In this letter, we proposed the use of feed-forward multilayer perceptron in deep learning-based artificial neural network (ANN) to accurately predict 12 optical parameters of silica-based photonic crystal fiber (PCF) within milliseconds using 6 input parameters. The optimized ANN has 3 hidden layers and each layer has 50 neurons. The PCF has several hexagonal-shaped layers with circular air holes, and it uses silica as the cladding and FK51A glass as the core. The PCF parameters that have been successfully predicted include birefringence, chromatic dispersion, effective area, effective refractive index, nonlinear coefficient, numerical aperture, power fraction, relative sensitivity, V-parameter, and loss profiles such as confinement loss, effective material loss, and scattering loss. The prediction has high accuracy with a loss of only 0.00567 and a learning rate of 0.0001. 7-fold validation and batching are used to increase scalability during validation. The proposed ANN is over 99.9% faster than conventional numerical simulation approaches.

Biomimicry offers natural and effective solutions to solve critical challenges in wide range of emerging technologies. This paper introduces several bio-inspired dynamic microwave photonic technologies as well as biomimicry in fiber-optic sensors enhanced soft robotics.

Groundwater level provides critical insight to public resource allocation and climate variability. Remote monitoring of groundwater level is demonstrated, based on wavelength-shift induced optical power change in fiber Bragg grating caused by water pressure fluctuations.