| Summary: | Pneumatic Artificial Muscles (PAMs) are a class of pneumatic drives that have received
considerable attention for applications related to bio-inspired robotics. Nevertheless, servo
control of PAMs is challenging due to the compressibility and nonlinear flow characteristics of
air, hysteresis behavior as well as uncertainties present. In this paper, positioning of an
antagonistically paired PAMs with mass flow rate of compressed air regulated by a 5/3-way
proportional directional valve and driven by a Dynamical Adaptive Backstepping-Sliding Mode
Control (DAB-SMC) scheme is investigated. Implemented for the first time on a PAM-driven
actuating system, derivation of this model-based nonlinear control scheme is presented first
followed by experimental evaluation. Positioning performance is studied using a sinusoidal
trajectory with tracking frequencies of 0.05Hz, 0.1Hz , 0.2Hz and 0.5Hz, and a multiple-step
polynomial input having step sizes of 0.7°, 1.4°, 2.9° and 5.7°. Over various operating conditions,
average root mean square error (RMSE) value of 0.16° and steady-state error (SSE) value of
0.04° are achieved for position tracking and regulating, respectively. The adaptive LuGre friction
observer embedded in the control scheme effectively compensates hysteresis behavior of the
PAMs and helps to improve the performance. The proposed DAB-SMC scheme outperforms the
classical sliding mode control scheme by 33% in accuracy. The control scheme is also
demonstrated a robust performance towards the uncertainties including loading. In addition, a
slight performance compromise in both tracking and regulating tasks was observed, when the
5/3-way proportional valve is replaced by cost-effective 2/2-way pulse width modulation
controlled on-off valves.
|
|---|