Editorial: Materials, design, modeling and control of soft robotic artificial muscles

Xiaonan Huang, Andrew P. Sabelhaus, Mohammad Khalid Jawed, Lihua Jin, Jun Zou, Yuzhen ChenFrontiers in Robotics and AI 2022 · DOI · OpenAlex

Abstract

This editorial provides a roadmap for the next generation of soft robotics, synthesizing research that bridges the gap between advanced functional materials and intelligent motion. By integrating bio-inspired design with machine learning and decentralized control, the featured studies aim to develop artificial muscles that rival biological systems in capability and efficiency.

Introduction

Soft robotics represents a paradigm shift from traditional rigid-link systems, offering inherent safety for human interaction and superior adaptability to unstructured environments. Despite these advantages, the field faces a "performance bottleneck" that prevents soft artificial muscles from reaching their full biological potential. This editorial highlights a collection of research addressing four critical gaps: material performance, hardware miniaturization, modeling of nonlinear mechanics, and high-fidelity control systems.

The four pillars of soft robotic artificial muscle research: Materials, Design, Modeling, and Control.

Methodological Innovations

The featured research synthesizes novel approaches across five distinct domains:

  • Chemical Engineering: Researchers explored "doping" polydimethylsiloxane (PDMS) with imidazole-based ionic liquids. This process creates high-permittivity elastomers suitable for sensitive, stretchable sensors.
  • Biomimetic Design: One study utilized superabsorbent polymer gels encapsulated in "bubbles" to create actuators that respond autonomously to environmental stimuli, such as changes in salinity.
  • Deep Learning for Hysteresis: To address the complex "memory" effect (hysteresis) in thermal actuators, Long Short-Term Memory (LSTM) neural networks were employed to predict bending angles with high precision.
  • Classical Mechanics: Advanced modeling techniques, including Euler-Bernoulli beam models and Discrete Elastic Rod models, were used to characterize the stiffness and damping properties of fluidic muscles.
  • Neuro-inspired Architecture: A theoretical framework inspired by the "octopus model" explores distributed control, where intelligence is delegated to the limbs rather than centralized in a single processor.

Key Insights and Quantitative Results

The research collection yields several significant findings that push the boundaries of soft actuation:

  • Enhanced Material Sensitivity: By incorporating chloropropyl silicone oil and ionic liquids into standard elastomers, researchers increased dielectric permittivity by 105 times. This allows for vastly more sensitive sensors that maintain a high stretchability of 321% and a low Young’s modulus (0.78 MPa).
  • Environmentally Responsive Actuation: The "soft bubble" actuators demonstrated the ability to react directly to their surroundings—swelling in fresh water and contracting in salt water—producing a maximum force of 10.96 N. This approach could eliminate the need for complex electronics in specific underwater applications.
  • Predictive Modeling: The LSTM framework successfully captured the core physics of electrothermally-actuated limbs, providing a robust path for their application in wearable robotics where traditional physics-based models often fail due to lag.
  • Dynamic Stiffness Control: For Extensile Fluidic Artificial Muscles (EFAM), increasing internal pressure from 25 to 100 psi was shown to increase bending stiffness by 400%, offering a mechanism for tunable mechanical properties.

Future Directions: Toward Autonomous Systems

While significant progress has been made in individual components—sensing, actuation, and modeling—the ultimate goal remains full system integration. Future research must prioritize:

  1. Untethered Operation: Reducing reliance on bulky external hardware (pumps, power supplies) to enable free-roaming soft robots.
  2. Closed-Loop Control: Developing feedback systems that seamlessly bridge new functional materials with the predictive power of AI.
  3. Decentralized Intelligence: Implementing architectures where soft robots can sense, process, and act locally, reducing computational overhead and mimicking biological efficiency.

By addressing these challenges, the field moves closer to creating autonomous soft systems capable of operating effectively in the complex, real-world environments for which they were designed.

Summary updated

Abstract

Published version
BibTeX
@inproceedings{huang2022editorial,
  author = {Xiaonan Huang and Andrew P. Sabelhaus and Mohammad Khalid Jawed and Lihua Jin and Jun Zou and Yuzhen Chen and M. Khalid Jawed},
  title = {{Editorial: Materials, design, modeling and control of soft robotic artificial muscles}},
  booktitle = {Frontiers in Robotics and AI},
  year = {2022},
  doi = {10.3389/frobt.2022.1074549},
}
Figure