Abstract

This research dissertation presents the design, fabrication, modeling and control of an artificial muscle (AM) actuated wrist joint system, i.e., a thermoelectric (TEM) antagonistically driven shape memory alloy (SMA) actuator, to mimic the muscle behavior of human beings. In the developed AM based wrist joint system, the SMA, exhibiting contraction and relaxation corresponding to its temperature, is utilized as the actuator in the AM. Similar to the nerve stimulation, TEM is introduced to provide heat stimulation to the SMA, which involves heating and cooling of the SMA.

SMA possesses superelastic behavior that provides a large force over its weight and effective strain in practical applications. However, such superior material has been underutilized due to its high nonlinear hysteresis behavior, strongly affected by the loading stress. Using the data obtained from the experiments, based on the Prandtl-Ishlinskii (PI) model, a Stress-Dependent Generalized Prandtl-Ishlinskii (SD-GPI) model is proposed, which can describe the hysteresis behavior of the SMA under the influence of various stresses. The parameters of the SD-GPI models at various stresses are obtained using a fitting function from the Matlab. The simulation results of the SD-GPI showed that prediction error is achieved at mean values of ±2% and a standard deviation of less than 7%. Meanwhile, the TEM model is also developed based on the heat balance theory. The model parameters are identified via experimental data using Range-Kutta fourth order integration equation and Matlab curve fitting function. The TEM model has shown a satisfactory temperature prediction. Then, by combining the obtained two models, an integrated model is developed to describe the whole dynamics of the wrist joint system.

To control the SMA actuated wrist system, the SD-GPI inverse hysteresis compensator is developed to mitigate the hysteresis effect. However, such a compensator shows errors in compensating the hysteresis effect. Therefore, the inverse hysteresis compensator error and the system tracking error are analyzed, and the adaptive back-stepping based control approach is adopted to develop the inverse based adaptive control for the antagonistic AM wrist joint. Subsequently, a corresponding control law is developed for the TEM system to generate the required temperature obtained from the adaptive controller. Simulations verified the developed approach. Finally, experiments are conducted to verify the proposed system. Input sinusoidal signal with frequency 0.1rad/s and amplitude of ±0.524rad (±30°) is applied to the wrist joint system. Experimental results verified that the TEMs antagonistically driven SMA actuators for artificial muscle resembling wrist joint has been successfully achieved.