Self organizing fuzzy control of pneumatic artificial muscle for active orthotic device

The design and development of an offline self organizing fuzzy controller for a pneumatic artificial muscle (PAM) is discussed in this paper. This inherently compliant soft actuator is intended as the driving component in a mobile, lower limb, assistive/rehabilitative orthotic device. Accurate control using classical control approach is difficult due to the compressibility of gas and the nonlinear elasticity of the PAM bladder. To overcome this, a self-organizing fuzzy controller is proposed and tested in simulation. Results for step, ramp and sinusoidal inputs exhibit excellent tracking capability. The rise time for a 17% contraction (knee contraction angle of 60° - upper limit of average knee contraction during walking) step input is 0.2s, while the steady state error is well below 2%.