Specklegram Multimodal-Deformation Sensing Using a Soft Optical Waveguide

Abstract

Multimodal deformation sensing is a key technology in healthcare monitoring, wearable devices, and robotics. However, significant challenges remain in developing devices capable of simultaneously sensing multiple types of stimuli, as such systems typically require complex structural designs or multiple sensing units, which limit their flexibility and scalability. Here, we propose a structurally simple and mechanically flexible sensor made of soft material, where complex deformations can be inferred by learning the output specklegrams. The simplified one sensing mechanism for multimodal deformation stems from the inherent speckle sensitivity of the soft optical waveguide with a decoding strategy based on deep learning. The proposed soft optical waveguide sensor is low-cost, flexible and can be fabricated by rapid 3D printing. By employing a deep convolutional neural network, the proposed deformation sensor accurately differentiates bending, stretching, twisting, and combined deformations, maintaining a high accuracy close to 100% over three days. Our multimodal deformation sensor promises advancements in multi-functional sensing systems for future applications.