Application Cases

Introduction

Soft actuators are becoming an important alternative to traditional motors and rigid structures in the field of humanoid facial robots, owing to their advantages of large deformation, high compliance, and lightweight properties, which can effectively address issues such as excessive mass, high noise, and stiff movements in existing facial robots. Although soft actuators based on pneumatic, shape memory alloy, twisted and coiled, and dielectric elastomer mechanisms have been applied to facial robots, they still face challenges including low integration density, complex control, slow response, or excessively high driving voltages. PVC gel electro-active actuators demonstrate promising application potential due to their fast response (<50 ms), low driving voltage (<0.4 kV), and strain and stress levels comparable to human muscle. However, existing multilayer PVC gel actuators suffer from limitations such as insufficient contraction strain (~10%), significant displacement decay after prolonged operation, empirically determined electrode structures and gel parameters, unclear effects of layer stacking on output characteristics, and a lack of standardized encapsulation modules, all of which constrain their integration and practical application in real robots. Therefore, systematic optimization of PVC gel material formulations and anode structural parameters, establishment of influence models for layer stacking on output performance, and design of standardized encapsulation modules with adjustable preload have become critical research directions for advancing the practical development of soft actuators toward humanoid facial robot applications.

This study represents a frontier branch in the field of soft robotics technology, specifically in the direction of electroactive polymer actuation, and is highly aligned with the core development goals of soft robots: "large deformation, high compliance, lightweight, and biomimetic motion." Addressing the common bottlenecks of insufficient output strain (traditionally ~10%), low standardization, and difficult system integration that currently exist in soft actuators, this study optimizes the PVC gel material formulation and mesh anode structural parameters, establishes a dynamic model of the effect of layer stacking on output performance, and designs a standardized encapsulation module with adjustable preload, increasing the actuator contraction strain to over 20% and the operating bandwidth to over 9 Hz. Simultaneously, this technology was systematically integrated into a humanoid facial robot for the first time, achieving smooth multi-degree-of-freedom facial expression movements and precise imitation of a singer's jaw motion, with the final robot weighing only 754 g and featuring simple control. This not only verifies the feasibility of PVC gel actuators as high-performance artificial muscles, but also provides a typical example for the development of soft robots from single-component research toward highly integrated, highly robust, multi-degree-of-freedom practical systems.

Research Direction

Polyvinyl chloride (PVC) gel electroactive soft actuator; dynamic displacement prediction model for multilayer actuators based on the Hill muscle model and the three-element phenomenological model; effect of plasticizer content on the mechanical properties and surface micromorphology of gels.

Experimental objective

The core flexible component of this facial expression robot is fabricated by laminating PVC gel electrostrictive flexible material with stainless steel mesh electrodes. Under the action of an electric field, the gel undergoes creeping deformation, causing it to fill into the mesh apertures of the electrodes under the electric field, thereby generating expansion and contraction motion in the thickness direction. It returns to its original state upon removal of the electric field. Through ON-OFF control, muscle-like expansion and contraction motion is achieved. Different forms of deformation motion can be realized by adjusting the waveform and frequency of the electric field signal. This process requires a power amplifier to amplify the electrical signal by 10–100 times, thereby driving each flexible component to undergo deformation and enabling the expression robot to produce various facial expressions.

Testing equipment

Scanning Electron Microscope (ZEISS Sigma 300),Electronic Universal Testing Machine (CMT6103 MTS),Advanced Rheometer (SmartPave 92),Laser Displacement Sensor (LK-H050, KEYENCE),High-Voltage Amplifier (ATA7030, Xi'an Aigtek Electronics),Oscilloscope (UTG926E),NI myRIO (1900, NI),Programmable Power Supply (D2105P-2000-250),Laser Scanning Confocal Microscope,Electronic Balance,DC Power Supply,Fixed Pulley and Weight Load Device,High-Speed Camera.

Experimental process

A test platform for the facial expression robot was constructed based on the ATA-7030 power amplifier. The experimental equipment included a signal generator, an ATA-7030 power amplifier, a laser displacement sensor, an oscilloscope, and others. Electrical signals of different waveforms and frequencies (-10 to 10 V) were sent from the signal generator to the ATA-7030 power amplifier, which amplified them by approximately 50 times to output a high voltage of around 500 V, thereby driving the flexible devices of each module of the facial expression robot to achieve various facial expression changes.

Figure1 Illustration of the expression robot based on PVC gel actuators. a) Robot model. b) Robot internal structure. c) PVC gel actuator module. d) Decomposition  diagram of the actuator module. e) Multilayered PVC gel actuator structure.

Figure2  experimental setup.

Experimental results

The results indicate that the ATA-7030 power amplifier can stably amplify signals of different types and frequencies, thereby enabling different deformations of the flexible actuators and ultimately achieving various facial expressions of the expression robot.

The results show that the flexible actuators before and after modular packaging exhibit equivalent driving characteristics, with the displacement reduction after packaging being only approximately 10%. Meanwhile, for different input signals, the output displacement variation of the actuator can satisfactorily follow the input voltage variation. After 10,000 cyclic driving cycles, the displacement loss of the flexible actuator is only approximately 5%, demonstrating favorable cyclic life. The results demonstrate that the ATA-7030 power amplifier can stably amplify input signals of different types and frequencies, featuring fast response, large amplification factor, and small output error. It can fulfill the signal processing requirements for non-standard customized input signals of flexible actuators, thereby enabling different deformations of the flexible actuators and ultimately achieving various facial expression changes of the expression robot.

Figure3 Design and characterization of M-MPGA. a) The module of MPGA. b) Comparison of displacements before and after MPGA packaging. c) Comparison of output force before and after #60-MPGA16 packaging. d) Comparison of output work before and after #60-MPGA16 encapsulation at 400 V. e) Frequency response Bode curve of M-MPGA. f) M-MPGA leveling test at 400 V with different preload factors. g) M-MPGA response variation at 1 Hz square wave. h) Response variation of the M-MPGA at 1 Hz triangle wave. i) Response variation of the M-MPGA at 1 Hz sine wave. j) Displacement distribution of the M-MPGA over 10000 actuation cycles (about 2.78 h) at 400 V, 1 Hz.

Figure4  The 7 basic expression voltage waveforms and expressions. a) Eyebrow voltage waveform. b) Eyelid voltage waveform. c) Jaw voltage waveform. d) Eyeball  voltage waveform. e) Seven basic expressions compared to the model.

The effectiveness of the amplifier in this experiment

1. Amplify low-voltage control signals into hundreds of volts of high voltage to drive the PVC gel actuator to produce controllable contraction displacement, achieving the motion of each robotic actuation mechanism.

2. Output high-voltage excitation signals of different frequencies and waveforms to test the frequency response characteristics, operating bandwidth, and response time of the actuator.

3. Cooperate with the laser displacement sensor to perform a 0–440 V voltage sweep, calibrate the voltage–displacement relationship of the actuator, and determine the theoretical voltage values required for each actuation mechanism.

4. Output corresponding voltage waveforms according to preset facial expressions and motion requirements, drive multi-module coordinated motion, and achieve smooth facial expression switching and dynamic behavior imitation.

5. Convert low-voltage control signals into high-voltage drive signals to meet the driving requirements of the actuator while ensuring the operational safety of the experimental and robotic systems.

Application fields

Humanoid service robots,Wearable medical rehabilitation devices,Flexible haptic feedback and sensing systems,Biomimetic soft robots.