TacCap: A Wearable FBG-Based Tactile Sensor for Seamless Human-to-Robot Skill Transfer

TL;DR

TacCap tackles the lack of tactile feedback in large-scale human demonstrations for robot skill transfer by introducing a wearable FBG-based tactile sensor that can be used on both human and robot fingertips. The approach features a lightweight, EMI-immune three-layer thimble housing a single optical fiber with distributed FBGs, enabling high-fidelity tactile data at 2 kHz sampling. The paper details design, fabrication, calibration, and a learning-based contact-prediction model, and demonstrates grasp stability prediction with a low human-to-robot transfer gap while open-sourcing hardware and software. Collectively, these contributions provide a practical path toward scalable, tactile-enabled human-to-robot skill transfer for dexterous manipulation in real-world settings.

Abstract

Tactile sensing is essential for dexterous manipulation, yet large-scale human demonstration datasets lack tactile feedback, limiting their effectiveness in skill transfer to robots. To address this, we introduce TacCap, a wearable Fiber Bragg Grating (FBG)-based tactile sensor designed for seamless human-to-robot transfer. TacCap is lightweight, durable, and immune to electromagnetic interference, making it ideal for real-world data collection. We detail its design and fabrication, evaluate its sensitivity, repeatability, and cross-sensor consistency, and assess its effectiveness through grasp stability prediction and ablation studies. Our results demonstrate that TacCap enables transferable tactile data collection, bridging the gap between human demonstrations and robotic execution. To support further research and development, we open-source our hardware design and software.

Figures (8)

• Figure 1: (a) A human operator wearing TacCap tactile sensors to collect demonstration data. (b) A robot hand equipped with the same sensors executing tasks with tactile feedback.
• Figure 2: (a) Schematic of the TacCap sensor design and its components. (b) Unwrapped projection of the sensor surface, showing FBG sensor locations.
• Figure 3: Finite element analysis of the tactile sensor illustrating the strain distribution across the dome structure. The color gradient represents strain magnitude, with red indicating regions of higher strain.
• Figure 4: Calibration setup for collecting contact prediction data.
• Figure 5: Illustration of the raw wavelength signal processing during calibration. A sliding window is applied to capture signal history. Signals at the right end of the sliding window, falling within the green region, are considered positive (indicating contact), while those in the red region are considered negative.