Designing Soft Arms with Octopus-Like Dexterity: Insights from Magnetic Resonance Imaging and Finite Element Analysis

Abstract

Octopuses are capable of remarkably intricate movements without a skeletal framework, making them a compelling model for the design of soft robotic arms. While previous research has explored the bending, elongation, and shortening of octopus arms, the spatial distribution of specific muscle groups along the arm and their functional implications remain underexplored. In this study, high-resolution magnetic resonance imaging of 24 arms from Octopus bimaculoides was used to quantify the distribution of transverse, aboral, oral, and lateral internal longitudinal muscles, as well as the axial core housing the nerve cord. Results revealed a progressive increase in axial core area and a decrease in transverse muscle area from proximal to distal arm regions, while longitudinal muscle distributions showed no consistent trend. These anatomical insights informed the design of four soft arm models. Two models incorporated either uniform or octopus-inspired muscle group distributions, and the other two included an additional passive axial core. Using silicone rubber to mimic muscle mechanics, each design was evaluated via finite element analysis for tip displacement and arm curvature across various motions. The bioinspired model without an axial core achieved the greatest tip displacement, while the inclusion of the core reduced performance. Moreover, a parametric analysis of transverse-assisted bending demonstrated that even modest changes in the activation levels of transverse and longitudinal muscles can produce markedly different arm curvatures. This highlights how a bioinspired architecture can enable complex movements through simple modulation of relative muscle activation. Together, these findings underscore the value of biologically informed design principles in advancing the dexterity and agility of next-generation soft robotic arms.