Impedance-Modulated Soft Strain Sensor with High Stability for Humanoid Robots.

Soft strain sensors are crucial for enabling humanoid robots to perform industrial, medical, and other human-related tasks. However, the limited internal space of humanoid robots exposes soft strain sensors to interference from line resistance, contact resistance, and alternating magnetic fields generated by motor actuator systems and power conversion circuits. To address this issue, inspired by biological neural signal systems, this work proposes an impedance-modulated soft strain sensor with high stability. The sensor combines a liquid metal (LM) resistor, a capacitor, and an inductor to form a passive band-stop filter, which is encapsulated in a soft elastomer. Tensile strain increases the resistance of the LM resistor, reducing the impedance of the sensor at resonance and converting the resistance signal into an impedance-modulated signal. Based on filter theory, the circuit structure of the sensor and the selection of the resistance, capacitance, and inductance components are analyzed in terms of stability, sensitivity, and measurement feasibility. By employing a series connection of resistance and inductance, the sensor achieves high impedance at resonance, effectively suppressing interference from line and contact resistance. Additionally, the frequency of the impedance-modulated sensor does not overlap with the frequency of electromagnetic interference (EMI) from the humanoid robot motor drivers, achieving immunity to EMI. Furthermore, a Field Programmable Gate Array-based signal acquisition system is constructed to measure the impedance-modulated signal. Finally, an application of this impedance-modulated sensor in humanoid robots and wearable devices is demonstrated.