Precise control over deformation behaviors is essential for shape-morphing devices, adaptive actuators, and soft robotics. However, designing materials to achieve the whole user-defined deformation pathway remains a significant challenge. This work introduces a fast, physics-informed approach, namely incremental contrastive learning, for the inverse design of Willis lattice metamaterials capable of achieving desired deformation pathways under prescribed external constraints. We first implement gauge transformation to analytically derive an initial Willis lattice configuration and then leverage contrastive learning to develop a local learning rule on each incremental deformation step so as to design strain-dependent Willis parameters to realize the desired deformation pathway. Numerical simulations are performed to validate the approach and demonstrate its ability to program diverse and complex target deformations prescribed on boundary nodes, internal nodes, or even all nodes. Notably, we demonstrate independent control over both the final deformed shape and the specific pathway taken to achieve it, such as programming “early” versus “delayed” convexity to reach an identical final configuration. The developed framework provides a robust and efficient solution for programming matter, opening new avenues for designing complex morphing structures and tunable systems.

Close

PrevNext

View figure in article

• Analog computation
• Elasticity
• Inverse problems
• Physics of computation
• Strain
• Actuating materials
• Mechanical metamaterials
• Machine learning