I. Introduction
Modern mechatronic systems, such as assembly line robots and surgical robots, often need to perform complex target motions depending on their specific tasks. To achieve this goal, many intelligent actuator mechanisms are developed. For example, an origami-inspired pneumatic actuator was proposed in [1] to realize robust elongation and contraction motions. Another origami-based multi-channel actuator was developed by authors in [2] to achieve flexible bending movements for gastrointestinal endoscopic applications. Other studies also used auxetic-inspired structures to realize crawling [3], bending [4] and twisting [5] motions for actuators. On the other hand, in order to achieve complex target pose with fewer actuation degrees of freedom (DOF), many researchers have encoded the deformation information into the initial geometry of the actuators [6]–[8].