Shape memory alloys are a class of smart materials that can recover their original shape by heating above a temperature called austenite transformation temperature when subjected to deformation at low temperature. This property enables the shape memory alloy to be used as a unique actuator. Also, the material’s resistance changes during transformation. Thus, the change in resistance and other electrical properties can be used to sense the deformation, which eliminates the requirement of additional sensors. An artificial neural network, described in our earlier study, was able to accurately model the relationship between the electrical properties and manipulator position. This study aims to develop a control methodology of a shape memory alloy–actuated rotary manipulator using the feedback signal from the previously developed artificial neural network model, thus eliminating the need of any external position sensor. The control methodology using variable structure control technique is experimentally tested under different conditions. Also, the effect of environmental temperature on the ability of artificial neural network to predict manipulator position is analyzed using phenomenological model simulations. It is concluded that this system gives a robust performance with a small tolerance (less than 5°) and can operate well even when the ambient temperature changes considerably.
