I have been working on the nonlinear control of Pneumatic Artificial Muscles (PAM) for the last 2 years. The contributions are;
1) A detailed model that encapsulates inherent PAM nonlinearities (friction, nonlinear elasticity) was constructed.
2) Using the dissipativity theory, we've synthesized a force controller controller with guaranteed stability. It can cope with intrinsic PAM nonlinearities. We also demonstrated its robustness.
3) The actuator redundancy in antagonistic setups was exploited by developing a mapping from Joint Torque & Stiffness to individual PAM forces.
In doing so, joint torque & stiffness control could be interpreted in terms of PAM force control with sensory feedback.
4) Position control, a challenging task in PAM-powered systems (unlike motors), was addressed and its stability was proven in the sense of Lyapunov.
5) The controller was also implemented for a bipedal exoskeleton system that is actuated via PAMs.
6) A balance controller which makes use of the variable physical stiffness was constructed.
Its effectiveness is demonstrated with respect to optimal constant stiffness.
Related publications are under review, and preprints will be added later.