High-Fidelity, Customizable Force Sensing for the Wearable Human-Robot Interface

TL;DR

Fluidic innervation introduces a high-fidelity, customizable force-sensing modality for wearable human–robot interfaces by embedding air-filled channels in a 3D-printed silicone pad and reading pressure with a differential transducer. Benchtop tests show a linear pad-force relationship with $R^2=0.998$, while in-human tests with a clinical dynamometer reveal strong torque–pressure correlations for above-knee flexion ($R^2=0.95$) and below-knee extension ($R^2=0.75$). The approach is validated during dynamic bicep curls and unpowered squats, where pad pressure tracks elbow angle and squat phase with high fidelity and minimal drift. These results position fluidic innervation as a rapid, adaptable sensing solution for real-time control and user-function assessment in wearable robotics.

Abstract

Mechanically characterizing the human-machine interface is essential to understanding user behavior and optimizing wearable robot performance. This interface has been challenging to sensorize due to manufacturing complexity and non-linear sensor responses. Here, we measure human limb-device interaction via fluidic innervation, creating a 3D-printed silicone pad with embedded air channels to measure forces. As forces are applied to the pad, the air channels compress, resulting in a pressure change measurable by off-the-shelf pressure transducers. We demonstrate in benchtop testing that pad pressure is highly linearly related to applied force ($R^2 = 0.998$). This is confirmed with clinical dynamometer correlations with isometric knee torque, where above-knee pressure was highly correlated with flexion torque ($R^2 = 0.95$), while below-knee pressure was highly correlated with extension torque ($R^2 = 0.75$). We build on these idealized settings to test pad performance in more unconstrained settings. We place the pad over \textit{biceps brachii} during cyclic curls and stepwise isometric holds, observing a correlation between pressure and elbow angle. Finally, we integrated the sensor into the strap of a lower-extremity robotic exoskeleton and recorded pad pressure during repeated squats with the device unpowered. Pad pressure tracked squat phase and overall task dynamics consistently. Overall, our preliminary results suggest fluidic innervation is a readily customizable sensing modality with high signal-to-noise ratio and temporal resolution for capturing human-machine mechanical interaction. In the long-term, this modality may provide an alternative real-time sensing input to control / optimize wearable robotic systems and to capture user function during device use.

Figures (7)

• Figure 1: Demonstration of a person wearing the sensorized pad at (A) rest and (B) full elbow flexion. The insets show representative data of the reported pad pressure. (C) Overview of the sensorized pad system. From left to right, the black fluidically innervated pad on a strap. The pad is connected by a white barb and transparent silicone tubing to an off-the-shelf pressure transducer on a green breadboard. The transducer connects to an ESP-32 microcontroller which streams the data to the main computer. All scale bars are 5
• Figure 2: CAD render and system overview of the fluidically innervated pad. (A) Render of the entire pad with upper slot for strap attachment. (B) Transparent view without the strap slot showing interconnected channel structure yielding one bulk pressure reading. (C) Schematic of connection to readout electronics. All openings are sealed (black) except one connected via tubing (blue) to a differential pressure transducer measuring pad pressure relative to ambient air, with data recorded by a microcontroller.
• Figure 3: Characterization of sensorized pad using mechanical testing machine. (A) Mean compression distance vs. force across three trials. The maximum standard deviation was 0.15 and $<10^{-2}$ (not visible). The inset shows the compression test setup (scale bar 1 ). (B) Mean force vs. sensorized pad pressure was highly linear ($R^2$ 0.998, maximum standard deviation 32 , not visible). (C) Exemplar trial in which the pad was compressed until 20 , after which the position was held for 2 minutes. Stress relaxation in pressure readings were observed due to viscoelastic material properties, with an exponential decay time constant $\tau$ of 26.6 .
• Figure 4: Characterization of sensorized pads through on-human dynamometer testing. Participants repeated isometric volitional knee torque output to 10 Nm while dynamometer torque and pad pressure were recorded simultaneously. (A) Exemplary time series data from knee flexion with (B) the fluidically innervated pad above the knee. (C-D) Correlation results between the measured pressure of the above-knee pad and the measured dynamometer torque during flexion and extension. (E) Tests were also performed with the fluidically innervated pad below the knee. (F-G) Correlation results between the measured pressure of the below-knee pad and the measured dynamometer torque during flexion and extension. Scale bars represent 5 .
• Figure 5: (A) Bicep curls from 90° to full elbow flexion back to 90° were repeated with varying load held in the hand ranging 0--4.54 (scale bar 5 ). (B) Pressure data normalized in time by cycle percentage is shown for each load (orange and black lines are mean± 1 standard deviation, respectively).