Data-driven tuning of nonlinear internal model controllers for pneumatic artificial muscles

This paper proposes a data-driven approach to control of the pneumatic artificial muscle (which is abbreviated as PAM). Although the PAM has a desirable advantage of flexible structures, the inherent nonlinear property makes it difficult to obtain a mathematical model and to design an appropriate controller. As one of the ways to overcome such difficulties, a data-driven controller tuning based on the direct use of the data is expected to be utilized. Particularly, we apply fictitious reference iterative tuning (which is abbreviated as FRIT), which is a data-driven controller tuning method that enables us to obtain a desirable controller parameter with only one-shot experimental data, for tuning of parameters of controllers to obtain desired response of the PAM. In addition, we show that it is possible to obtain not only the desired controller but also the mathematical model of the PAM by using the internal model controller (which is abbreviated as IMC) including a nonlinear mathematical model of the PAM. Finally, we give an experimental result in order to show the validity of the proposed method.