Multi-SpeckleForce: An Experimental Multimode Fiber Specklegram Dataset for Multi-Position Force Sensing

Abstract

This experimental dataset presents optical fiber specklegram images obtained from a step-index multi-mode fiber (MMF) due to multi-position force perturbations. The force recognition specklegrams can help in the development of data-driven models for simultaneous force estimation at multiple sensing locations along a single MMF. The dataset comprises 11,000 grayscale images with a resolution of 256 × 256 pixels, which belong to 550 records across five independent acquisition sets. Each record corresponds to a fixed triplet of applied transverse forces at three spatially separated sensing positions. The applied force values span a range from 0.26 N to 1.26 N and measured using digital force gauges with a readout resolution of 0.01 N. Key acquisition parameters include a 633 nm HeNe laser source, a 50 μm core multimode fiber of 2 m length, and CCD-camera-based specklegram imaging. Structured labels are provided in physical units of Newtons for all images. This dataset supports the development and systematic evaluation of multi-output regression models for specklegram-based force sensing and can facilitate comparative studies across experimental and data-driven approaches. Two baseline machine learning implementations with k-nearest neighbors (kNN) and Random Forest (RF) are provided to demonstrate the usability of the dataset for data-driven multi-position force estimation. Furthermore, this resource will enable research in applications such as robotic tactile sensing, structural health monitoring, and distributed force measurement systems.