Abstract—This paper presents a technique to control a robotic hand via a remote sensory glove. Five flex sensors are attached to the glove to capture the hand gesture which is then used to remotely control the robotic hand. Due to uncertain behaviour of the signals when sensors are bent, a structured controller consisting of two types of filters and a constraint block has been designed for compensation. The compensator was tested via an experiment with a controlled grab and release hand movement, and the performance is evaluated in terms of the speed when the fingers are bending or flexing, and the mismatch between the actual and expected sensor values when the fingers are stationary. Experimental results show that, with the proposed compensator, the speed is driven closer to the desired value, and the mismatch can be significantly reduced by approximately 45% when the fingers are in the grab position, and 96% when the fingers are totally released.