Lattice structure musculoskeletal robots: Harnessing programmable geometric topology and anisotropy

This is the authors' abstract. We don't add key points for this paper.

Natural musculoskeletal systems combine soft tissues and rigid structures to achieve diverse mechanical behaviors that are both adaptive and precise. Inspired by these systems, we propose a programming method for designing bioinspired soft-rigid robotic structures using lattice geometries made from a single material. By introducing previously unknown approaches to the geometric design of unit cells within lattice structures-based on continuous blending and superposition of existing lattice geometries-we can precisely tune stiffness and anisotropy. These designs enable the creation of three-dimensional structures with spatially varying mechanical properties, ranging from tissue-like compliance to rigid, bone-like load-bearing capabilities. Using these methods, we fabricated a musculoskeletal-inspired tendon-driven robotic elephant that integrates joints with programmable bending profiles, achieving a continuously soft trunk. Our lattice geometry generation techniques allow for over 1 million discrete configurations and infinite geometric variations, offering a scalable solution for designing lightweight, adaptable robots.