Design of Biomimicry robotic eye using thin Mckibben actuators with agonist-antagonist muscles

Abstract

The human eye is important to the survival and evolution of humanity. Thus, we look to it for inspiration in designing and translating its design into a robotic system that can be actuated by soft actuators. The purpose of this study is to investigate and implement biomimicry of the human eye through a 3D printed robotic eye based on a ball joint system that is actuated by thin McKibben actuators with agonist and antagonist muscles. This study proposes the implementation of the agonist-antagonistic as well as neutralizer muscles pairs to actuate the robotic eye through the control of air pressure at 200 kPa and 420 kPa investigate the behaviour of the thin McKibben actuators during the implementation of the agonist-antagonist muscles as well as the assessment of its performance via the laser pointer experiment and Tracker application. The solution is based on the advantage gained through the implementation of the thin McKibben actuators as well as 3D printing during the robotic eye design, which grants an improved, simpler design to the robotic eye. The approach has several notable merits, namely a reduction in design complexity and reduction in size. A comprehensive verification via experimentation was carried out to determine the effectiveness of the concept and design. The result confirms that the method can produce a robotic eye design that works well through the use of agonist-antagonist muscles pairs and neutralizer muscle pairs. From the experimentation works, it was found that the 3D printed robotic eye can be actuated via the Thin McKibben actuator with the angular movement of around 20 degrees and with two Degrees of Freedom via Agonist-Antagonist muscle pairing. The results also show the high repeatability of the robot while operating under hysteresis mode, with a standard deviation of 0.1 to 0.54 after repeated testing. Furthermore, the behaviour of the thin McKibben actuators during the implementation of agonist-antagonist muscle action was successfully identified and categorized as contraction, relaxation and re-contraction behaviours. The method proposed in this paper can be implemented in other types of robots with thin McKibben muscles as actuators as well.