Spatially graded stiffness is a strategy widely found in nature to guide asymmetric deformations, as observed in plant stems and tendrils through a process known as lignification. Based on this principle, modular, tendon-driven cylinders composed of 3D metamaterials with varying internal architectures are introduced. By locally adjusting the relative density across regions of each cylinder, distinct deformation modes, including bending, twisting, double bending, and contraction, are achieved using a singular tendon input. To determine the optimum strut thickness variation, computational and experimental compressive tests were performed. The cylinders are fabricated from elastic resin and incorporate embedded magnets, facilitating tool-free modular assembly into serial and parallel robotic configurations. The deformation behavior of multiple architectural patterns is characterized both experimentally and computationally. Additionally, the performance of various multi-module configurations is evaluated, including a hybrid assembly performing a simple grasping task. This approach enables motion to be physically encoded into the robot¿s structure, offering a scalable framework for programmable, minimally-actuated soft robotic systems.
