Abstract:Application of machine learning techniques on fiber speckle images to infer fiber deformation allows the use of an unmodified multimode fiber to act as a shape sensor. This approach eliminates the need for complex fiber design or construction (e.g., Bragg gratings and time-of-flight). Prior work in shape determination using neural networks trained on a finite number of possible fiber shapes (formulated as a classification task), or trained on a few continuous degrees of freedom, has been limited to reconstruction of fiber shapes only one bend at a time. Furthermore, generalization to shapes that were not used in training is challenging. Our innovative approach improves generalization capabilities, using computer vision-assisted parameterization of the actual fiber shape to provide a ground truth, and multiple specklegrams per fiber shape obtained by controlling the input field. Results from experimenting with several neural network architectures, shape parameterization, number of inputs, and specklegram resolution show that fiber shapes with multiple bends can be accurately predicted. Our approach is able to generalize to new shapes that were not in the training set. This approach of end-to-end training on parameterized ground truth opens new avenues for fiber-optic sensor applications. We publish the datasets used for training and validation, as well as an out-of-distribution (OOD) test set, and encourage interested readers to access these datasets for their own model development.

Deep learning based optical curvature sensor through specklegram detection of multimode fiber

Bending recognition based on the analysis of fiber specklegrams using deep learning

Integrated All-Fiber Sensor for Simultaneous Measurements of Curvature and Temperature

Learning to sense three-dimensional shape deformation of a single multimode fiber

Deep Learning-Based Simultaneous Temperature- and Curvature-Sensitive Scatterplot Recognition

Distributed curvature sensing using long period fiber grating and machine learning numerical analysis

Strain Sensor Based on Gourd-Shaped Single-Mode-multimode-single-mode Hybrid Optical Fibre Structure.

Three-Axis Curvature Sensor Based on Fiber SPR

High Sensitivity Curvature Sensor with Intensity Demodulation Based on Single-Mode-Tapered Multimode-Single-Mode Fiber

A Fiber Sensor for Different Curvature Measurement Based on Nano-Erbium-ytterbium Co-Doped Fiber Cascade Structure As Both Source and Sensor

An Inline Fiber Curvature Sensor Based on Eccentric Core Fiber and Off-Axis Air Cavity Fabry-Pérot Interferometer

Multimode optical fiber transmission with a deep learning network

Multimode Optical Fiber Specklegram Pressure Sensor Using Deep Learning

A novel method of curvature demodulation for Mach-Zehnder interferometer optical fiber curvature sensors

Distributed Temperature and Curvature Sensing Based on Raman Scattering in Few-Mode Fiber

A dual-parameter sensing system for temperature and curvature based on Nano-EYDF and in-line cascaded Mach-Zehnder interferometer

The Curvature Sensor Based on Fiber-Optic Spindle Arrays

Temperature-insensitive curvature sensor with few-mode-fiber based hybrid structure

Fiber-Optic Shape Sensing Using Neural Networks Operating on Multispecklegrams

Spatially Variant Deblur and Image Enhancement in a Single Multimode Fiber Imaged by Deep Learning

Composed multicore fiber structure for direction-sensitive curvature monitoring