Customizable bistable units for soft-rigid grippers enable handling of multi-feature objects via data-driven design

Abstract

Soft-rigid grippers represent a novel paradigm for grasping complex objects, combining the high deformability of soft components with the high stiffness of rigid components. Recently, bistable structures, as architected materials for engineering soft components, have attracted significant attention for their ability to enable rapid-response grasping and shape self-locking. However, challenges persist in utilizing them for handling multi-feature objects, i.e., irregular-shaped, fragile, and variable-weight objects. Here, we report a class of soft-rigid grippers comprising customizable bistable units and their data-driven design framework to address these challenges. Specifically, the transition behavior of bistable units can be tailored by designing their contact blocks (CBs), enabling grasping-force control of grippers for objects with varying fragility and weight. The CB design is achieved through an inverse design framework that employs extremely randomized trees (ERT) models and differential evolution (DE) algorithms. The trained ERT model accounts for the strongly coupled nonlinearity of structural deformation, material constitutive models, and contact behaviors during transition processes, achieving a prediction accuracy of 96.4%. Additionally, the grippers offer overload protection and shape-conforming reconfiguration for irregular-shaped objects. This bistable unit design offers grippers new ways of grasping complex objects, promising superb flexibility, scalability, and efficiency in the design and operation of robot technologies.