Predictive Fuzzy Model for Control of an Artificial Muscle

Modeling and control of a special type of the pneumatic actuator, called the Artificial Pneumatic Muscle (APM), is considered here. APM is able to exhibit many of the properties found in the real biological muscle. Contrary to its simple design, the control of APM is very complex due to its nonlinearity, compressibility of air, time varying properties, and difficulties in analytical modeling. In particular, the hysteretic behavior, which originates mainly from the extensive friction between different APM layers, is difficult to model. In order to overcome this problem, we propose a simple adaptive control law based on the APM predictive fuzzy dynamic model, which is able to cope with the inherent hysteresis of APM. Structure and parameters of the APM predictive fuzzy model are described through the fuzzy clustering of experimental data. The proposed predictive fuzzy model and its control law are verified through computer simulation.