Control-Oriented Models for Hyperelastic Soft Robots Through Differential Geometry of Curves

Area of Science:

• Robotics
• Materials Science
• Control Theory

Background:

• Soft robots require accurate, computationally efficient models for control.
• Existing models often lack detailed material behavior integration.
• Bridging material and control research is crucial for soft robot advancement.

Purpose of the Study:

• To develop a dynamic model for soft robots incorporating hyperelastic and viscoelastic material properties.
• To enable real-time simulations and enhance control strategies for soft robots.
• To validate the model and controller through simulations and experimental data.

Main Methods:

• Derived continuum dynamics using differential geometry of spatial curves.
• Integrated finite-element data for geometric and material nonlinearities.
• Introduced a reduced-order integration scheme for efficient dynamic Lagrangian matrix computation.
• Developed a passivity-based adaptive controller leveraging hyperelastic model parameterization.

Main Results:

• Achieved real-time (multilink) soft robot models with high numerical precision.
• Demonstrated enhanced robustness of the controller against material uncertainty and unmodeled dynamics.
• Validated the dynamic model with an additively manufactured soft robot manipulator under various conditions.
• Showcased good correspondence between model predictions and experimental results.

Conclusions:

• The proposed framework accurately captures soft robot dynamics with complex material behaviors.
• The passivity-based adaptive controller offers robust performance for real-time applications.
• The study provides a solid foundation for advanced control of soft robots through integrated modeling and experimentation.