Soft Robotic Joints With Anisotropic Stiffness by Multiobjective Topology Optimization

摘要

Soft robots inherit several merits associated with compliant mechanisms, including frictionless, monolithic fabrication, and backlash-free operation. Moreover, the incorporation of soft actuation bestows upon soft robots distinctive advantages in terms of achieving active mobility. However, current soft robot design usually focuses on the motion range in the pursued direction only, lacking of deliberate consideration of stiffness characteristics in other directions to withstand unexpected disturbances, which has greatly limited the use of soft robots in practical scenarios. In this article, we propose novel soft robotic joints with anisotropic stiffness characteristics that produce the desired active target motion upon actuation and, in the meantime, constrain the undesired deformation under disturbances. Such functionality and robustness are embodied in the delicate design of the joints' compliant structure, with a multiobjective topology optimization approach. We develop three basic soft joints, i.e., elongation, twisting, and omnidirectional bending, and the design process is automated by a unified computation framework. The optimized soft joint prototypes are quantitatively characterized in terms of the free motion, blocking force, and multidirectional stiffness. The experiment results show that the joints not only generate practicable active motions but equally importantly, attain greatly improved stiffness to withstand unexpected disturbances. We further showcase a dexterous wrist-hand system by assembling the soft joints and a soft gripper to navigate and manipulate objects in confined space, and develop cable-driven soft fingers with enhanced lateral and twisting stiffness for stable grasping.