Shape Matching: Evolving Fiber Constraints on a Pneumatic Bilayer

Bilige Yang, Joshua P. Powers, Atoosa Parsa, Josh C. Bongard, Rebecca Kramer-Bottiglio

Abstract: Shape-changing robots that can adapt their physical structure to changing tasks and environments require an ability to locally change their surface curvatures on-demand. Thin shape-shifting sheets represent one approach, and examples using differential growth have shown stimuli-responsive sheets that morph from one shape to another. In this paper, we present pneumatically-actuated bilayer sheets that can morph between shapes using surface fibers that constrain normal expansion while biasing in-plane expansion. The fibers can adhere to the pneumatic sheet to govern its shape trajectory, and can be easily detached and re-attached in various patterns to reconfigure the shape trajectory. We use artificial evolution to solve the geometric inverse problem of determining the fiber placements to attain a target shape upon inflation in simulation. For simple shapes, the evolved solutions converge toward a similar fiber placement strategy. For more complex shapes, evolution finds more diverse fiber patterns, which is desirable since some fiber patterns are more easily fabricated than others. We demonstrate the utility of evolving the fiber patterns in simulation by translating two shapes, a simple saddle and a monkey saddle, from simulation to reality.