Contents I. Introduction
Soft robots, primarily made out of intrinsically soft and/or flexible materials, are able to operate in clustered or unstructured environments and absorb energy arising from collisions [1], [2]. Rather than relying on translational and/or rotational joints to achieve a desired locomotion in the conventional rigid-bodied robots, a soft robot acquires mobility from the deformation of its elastic body [3]. Correspondingly, the actuation strategy is changed from single degree of freedom (DOF) motors to infinite DOFs dielectric elastomers [4], shape memory alloys [5], cable wires [6]–[8], chemical reaction [9], and pneumatic/hydraulic approaches [11]–[14]. Although the proliferation of worldwide research works on soft robots have demonstrated their applications on locomotive devices [10], [11], manipulation and human-serving tasks [12]–[14], they are still challenged by exerting high payloads and supporting their own weights. Looking at the anatomical structures of molluscs and vertebrates, one may observe that flexible tendons and hard skeletons are merged with soft tissues. This biological evidence suggests a similar solution in soft robots, which is combining soft materials with relatively harder ones to overcome the challenges faced by current soft robots [2]. To automatically design multimaterial soft robots with desired functionalities, a systemic framework is required because any changes in either the structure or material distribution will result in unpredictable changes.
