Maximizing Anthropomorphic Grasping Abilities of Bio-inspired Underactuated Robotic Hands

Recent neuroscience discoveries on human hand synergies have inspired the development of underactuated robotic hands, which replicate human-like grasping capabilities using a minimal number of actuators. However, a generalized methodology for determining the parameters of such bio-inspired underactuated hands to maximize anthropomorphic grasping abilities remains a significant challenge. To address this, we propose a novel framework based on Hertz contact theory to establish a general underactuated grasping model. Within this framework, we introduce evaluation indices and constraint conditions integrating morphological parameter ranges of the human hand derived from a scientific analysis in our prior work and an approximation index between human hand motions and robotic hand motions, aimed at: 1) biomimetic part: ensuring that the robotic hand's morphology, motion, and posture closely mimic those of the human hand, and 2) robotic part: maximizing the Euclidean norms of normal contact forces between the robotic hand and the object during grasping. To streamline the parameter optimization process, we devise a comprehensive, step-by-step strategy that groups parameters sequentially, enabling rapid convergence to optimal solutions. As a case study, we design and develop a dual-actuated robotic hand, comparing unaltered and optimized parameter schemes through extensive simulations and experimental validations. The results demonstrate the effectiveness of our method and suggest its potential applicability to a wide range of underactuated robots and bionic systems. This work provides a systematic approach to advancing the design and optimization of anthropomorphic robotic hands, bridging the gap between biological inspiration and engineering implementation.