Viscoelasticity-Based Mechanistic Modeling and Control of Bending Pneumatic Muscles
Zishuo Zhao, Baoguo Xu, Jiajin Wang, Jianwei Lai, Yifei Wang, Huijun Li, Aiguo Song
AI summary
Problem
Modeling bending pneumatic muscles is hindered by pronounced hysteresis and creep, while existing phenomenological and data-driven models lack physical interpretability or incur prohibitive computational costs for real-time control.
Approach
The authors derive a mechanistic model grounded in viscoelastic constitutive relations to capture material history dependence, then accelerate it using a sliding-window prediction mechanism to enable efficient feedforward-feedback hybrid control.
Key results
- VBMM achieves predictive accuracy with error below 3.69%
- Sliding-window optimization reduces computational overhead by 97%
- Feedforward-feedback hybrid control system successfully tracks BPM bending trajectories
- Linear viscoelastic framework validated for 15–40% strain ranges via stress relaxation testing
Why it matters
Provides a physically interpretable, computationally efficient modeling and control framework for soft pneumatic actuators, advancing their deployment in robotics and rehabilitation.
Abstract
Predicting the kinematics of bending pneumatic mus- cles (BPMs) remains challenging due to the necessity for mod- els that effectively address the pronounced hysteresis and creep inherent in soft materials. While prior research has predomi- nantly focused on phenomenological and data-driven modeling approaches, this study introduces a viscoelasticity-based mecha- nistic model (VBMM) and a feedforward-feedback hybrid con- trol system tailored for BPMs. First, the VBMM is developed by leveraging the principles of viscoelasticity—a common prop- erty of soft materials and a mechanistic driver of hysteresis and creep. Second, we address the computational challenge arising from the history-dependent viscoelastic response of BPMs, where the current state depends on the cumulative stress-strain history. Conventional methods incur escalating computational costs over time, rendering real-time control impractical. To resolve this, we propose a sliding window-based long-term prediction mechanism (long-term VBMM) that maintains model accuracy while signifi- cantly reducing computational overhead. Finally, a hybrid control system integrating the long-term VBMM as a feedforward compen- sator with feedback correction is designed to achieve BPM motion tracking. Experimental validation confirms the VBMM’s superior predictive accuracy (error < 3.69%) and demonstrates the control system’s effectiveness.