Position Control of McKibben-Type Pneumatic Artificial Muscle via Port-Hamiltonian Approach

Abstract

Pneumatic artificial muscles (PAMs) are soft actuators with softness and high power ratio. These characteristics provide a robot with adaptiveness and agility. However, controlling PAMs is difficult because they exhibit a highly nonlinear nature and involve complex multi-physical systems (i.e. mechanical and fluid systems). In particular, to achieve better control performance, the conventional approach suffers from estimating difficult-to-estimate parameters, such as the viscoelasticity of the PAM. To overcome the difficulty of controlling PAMs, this study utilized an interconnection and damping assignment passivity-based control (IDA-PBC) method via port-Hamiltonian (PH) framework. PH framework is well-suited for modeling PAMs because this framework describes the energy exchanges between the multi-physical systems of PMAs. In addition, the IDA-PBC method is a passivity based control that stabilizes the nonlinear systems. Loosely speaking, the feedback gain of the IDA-PBC method must be determined such that the PAM system is passive – that is, it either dissipates or conserves energy. Therefore, the PAM system can be stabilized without the precise viscoelastic parameters when the feedback gains are significantly higher than these parameters and ensure the system's passivity. Several experiments compared our proposed control with comparison control methods. As a result, our proposed control followed the desired position with a smaller error than the other comparison methods across a wide range of target positions.