Inverse Design of Planar Clamped-Free Elastic Rods from Noisy Data

The paper studies how to design the “natural shape” of a suspended 2D planar elastic rod so that, after deformation, it matches a chosen target shape. It first builds a theoretical framework from planar rod statics to compute the natural shape from the target. The authors then examine how uncertainties in the input data (such as noise) affect the accuracy of this inverse design. Because the theoretical approach struggles with noisy data, the paper introduces a learning framework that combines the statics model with an adjoint method for parameter sensitivity analysis. The framework is validated numerically for accuracy and robustness, and is positioned as useful for inverse design of soft structures, including soft robotics and morphing-structure animation.