Session: Research Posters

Paper Number: 165069

Autonomous Discovery: Tailored Near-Zero Poisson’s Ratio Metamaterials in Nonlinear Soft Gripper for Improved Interface for Interface Force Enhancement 

Advancements in computing, inference algorithms, and digital twin technologies have enabled autonomous discovery frameworks that surpass traditional human-driven design methods. These frameworks integrate model generation, simulation, and optimization algorithms to systematically explore and refine designs without direct human intervention. Such approaches have improved efficiency and performance in material design, chemical synthesis, and mechanical structure optimization. Robotic manipulation, particularly object gripping, is essential in industrial automation, enabling precise and repeatable operations in manufacturing, logistics, and assembly. Soft robotic grippers offer advantages over rigid grippers due to their compliance and adaptability, making them ideal for handling delicate or irregularly shaped objects. Pneumatic-driven soft grippers are widely used for their flexibility, lightweight structure, and cost efficiency. However, their payload capacity remains a challenge, as gripping force depends on the pneumatic chamber's ability to convert air pressure into mechanical force. While structural modifications such as bio-inspired architectures and fiber-reinforced designs have been explored, most rely on empirical methods, limiting optimization efficiency. Mechanical metamaterials (MMs) provide a promising solution to enhance soft robotic gripper performance. Engineered unit-cell architectures allow unique mechanical properties such as negative Poisson’s ratio (NPR) and near-zero Poisson’s ratio (N-ZPR), offering precise control over deformation and force transmission. Nevertheless, integrating data-driven MM structures into robotic manipulation has often required manual processes for realizing desired actuation behavior due to insufficient consideration of robotic manipulation scenarios.

Soft robotic systems utilizing nonlinear hyperelastic materials have transformed robotics by providing exceptional adaptability. However, designing structures that align with the complex behavior of nonlinear soft robots requires a method beyond intuition and manual adjustments. In this study, we introduce an Autonomous Design Discovery (ADD) framework that integrates Computer-Aided Design (CAD), Finite Element Method (FEM) simulation, and machine learning (Bayesian optimization) to enhance the gripping force of pneumatic-driven soft grippers. Beyond personal assumption that thinner chamber walls enhance gripping force by allowing greater expansion, the ADD framework identifies an optimal auxetic design with selectively thickened walls, exhibiting near-zero Poisson’s ratio (N-ZPR), a characteristic of mechanical metamaterials. This N-ZPR effect emerges from the combination of auxetic and conventional structures, which possess negative and positive Poisson’s ratios, respectively. The resulting deformation pattern minimizes transverse contraction while maximizing force transmission at chamber contacts and the fingertip-object interface. The optimized finger module demonstrated a 38% increase in interface force, validated through both simulations and experiments. Built upon this design, the gripper effectively manipulated objects up to 200 mm in height and 2,000 g in weight, successfully executing pick-and-place tasks using a 6-axis robotic arm. This study highlights a data-driven approach to designing robotic structures with nonlinear mechanical properties, leveraging physical inference and metamaterial principles to enhance robotic actuation efficiency.

Presenting Author: Baekgyu Kim Pusan national university

Presenting Author Biography: BaekGyu kim is currently pursuing a Ph.D-M.S. intergrated course under the supervision of Prof. Sang Min Park, Pusan National University. He started his academic studies in the field of Mechanical Engineering and received B.S. degree from Pusan National University, South Korea in 2020. His research interest is based on data-driven design, metamaterials, and self-powered sensors.

Authors:

Baekgyu Kim Pusan national university
Sang Min Park Pusan National University