Fabric Sensing of Intrinsic Hand Muscle Activity

TL;DR

This work tackles the challenge of inferring user intent for hand-assisted wearable robotics by targeting intrinsic hand muscles, which are often under-sensed by traditional forearm sEMG. It introduces a textile, three-channel sleeve that records sEMG from the thenar eminence and evaluates its ability to detect thumb movements and support open/close gesture classification, alongside a forearm armband for comparison. Key findings show that intrinsic-thumb activity yields large, distinguishable EMG signals (especially for thumb abduction) and can achieve competitive intent inference when fused with extrinsic sensing; performance is strong in healthy participants but more limited in a single stroke participant, highlighting the potential and current limitations for clinical translation. Overall, the textile-sensor approach offers a low-cost, non-obtrusive sensing modality for wearable hand robotics and rehabilitation, enabling improved intent inference and potentially better therapy outcomes.

Abstract

Wearable robotics have the capacity to assist stroke survivors in assisting and rehabilitating hand function. Many devices that use surface electromyographic (sEMG) for control rely on extrinsic muscle signals, since sEMG sensors are relatively easy to place on the forearm without interfering with hand activity. In this work, we target the intrinsic muscles of the thumb, which are superficial to the skin and thus potentially more accessible via sEMG sensing. However, traditional, rigid electrodes can not be placed on the hand without adding bulk and affecting hand functionality. We thus present a novel sensing sleeve that uses textile electrodes to measure sEMG activity of intrinsic thumb muscles. We evaluate the sleeve's performance on detecting thumb movements and muscle activity during both isolated and isometric muscle contractions of the thumb and fingers. This work highlights the potential of textile-based sensors as a low-cost, lightweight, and non-obtrusive alternative to conventional sEMG sensors for wearable robotics.

Figures (4)

• Figure 1: The fabric sensing sleeve. After embroidering the fabric electrodes and conductive thread, which act as wires to connect the sleeve to the OpenBCI board, the sleeve is assembled into a wearable form. The sensor array is positioned on the thenar eminence of the thumb to detect when the thumb moves into abduction.
• Figure 2: The 3-channel textile sensing array, labeled by color and channel number. Each textile electrode is made of conductive fabric embroidered to a fabric substrate with internal padding to improve skin-electrode contact. Conductive thread is stitched around the perimeter of the electrode to connect the electrode to the bioamplifier board.
• Figure 3: Schematic of experimental protocol. During the isolated movement task, participants moved their thumb or fingers from rest to an engaged position as shown. The isometric muscle contraction task required participants to resist external force from the experimenter's hand to engage the relevant muscles for the movement. Classification data collection involved participants opening and closing their hands into a lumbrical grasp, as shown.
• Figure 4: (best seen in color) Data from our fabric sensing sleeve and commercial armband (Myo) during the isolated muscle movement and isometric muscle contraction tasks over time (s). Grey bars denote isolated movement, and purple bars denote isometric contraction. Each row depicts data from one healthy participant, and each column corresponds to a hand movement (Fig. \ref{['fig:protocol']}). For each figure, the top plot shows the fabric sensor activity in $\mu V_{rms}$ (blue = Channel 1, orange = Ch. 2, green = Ch. 3). Y-axis limit is $500\ \mu V$ except for H6 (2,500 $\mu V$), and all cut-off peaks are below 1,000 $\mu V$. Bottom plot depicts corresponding Myo sensor activity in proprietary values (a.u.).