中国科学院自动化研究所机构知识库

In this article, a novel underactuated finger exoskeleton is designed to assist grasping tasks for the elderly with weak muscle strength. In mechanical design, the human finger's phalanges and joints are considered as part of the kinematic chains to realize the human-robot kinematic compatibility. The proposed finger exoskeleton achieves the finger preshaping and grasps objects with generic shapes. The proposed exoskeleton is both actively and passively backdriveable. Moreover, the weight of the wearable part of the proposed exoskeleton is 127 g, and the overall weight is 476 g, which indicates the proposed exoskeleton is lightweight and portable. To improve the grasping performance, the multiobjective genetic algorithm is implemented to optimize contact forces, which is to maximize the sum of the forces exerted on the index finger phalanges by the exoskeleton and to minimize the difference between the contact forces. After optimization, the sum of contact forces is risen by 15%, and the difference between forces is decreased by 53%. Furthermore, the admittance control is applied to make the proposed finger exoskeleton more compliant in the preshaping phase, and the admittance control is also implemented to achieve the fingertip grasping force control in the grasping phase. Finally, experiments have been conducted to verify the range of motion, grasping forces, and feasibility of the proposed index finger exoskeleton. The effectiveness of the control algorithm has also been verified by experiments.