Nonparametric control algorithms for a pneumatic artificial muscle

Accurate control of the PAM is a challenging task due to the nonlinear elasticity and the hysteresis losses in the actuator, the compressibility of air, and the nonlinearities in the pneumatic system. A novel implementation of a self-organizing fuzzy controller (SOFC) for the position and force control of a single PAM is proposed in this work. The performance of the controller is experimentally evaluated alongside two other nonparametric control algorithms; PD and fixed rule fuzzy controller (FFC). In addition a comparison of the performance of two different pneumatic valves (high speed on–off valve (HSV) and proportional pressure regulator (PPR)) is also undertaken to determine the influence of the valve type on the system response. In all the experiments for both position and force control strategies, the SOFC surpasses the PD and FCC in terms of percentage overshoot for step input and percentage RMS error (%RMSE) for ramp input whilst maintaining comparably rise time (Tr) and percentage steady state error (%SSE). It is able to track the reference signal with no oscillation and overshoot of less than 7% for both position and force control. Results show that the HSV performs better in response to abrupt changes (step input) while the PPR performs better in response to an incremental change (ramp input). 2012 Elsevier Ltd. All rights reserved.