In this paper,we propose an internal model control strategy with echo state networks considering the high nonlinearity of Pneumatic Artificial Muscle (PAM) and the high challenge of modeling it. We use two Echo State Networks to approximate the plant model and plant inverse model for constructing our Internal Model Control system. Recursive Least Squares (RLS) is used to train the nets.The performance of our method is tested by simulations on Matlab/simulink. The results show that satisfactory tracking perf...