A comprehensive review of dexterous robotic hands: design, implementation, and evaluation

Abstract

Dexterous robotic hands have been a central focus in robotics research, aiming to replicate the versatility and functionality of the human hand. This review provides a comprehensive analysis of the latest advancements in the literature on dexterous robotic hands, covering both hardware designs and implementation methods. We categorize robotic hand dexterity into potential dexterity, grasp dexterity, and manipulation dexterity, offering a systematic framework for evaluating robotic hand performance. Various dexterous hands are then organized based on their number of digits, transmission mechanisms, actuation methods, and sensing technologies, with their dexterity compared using different evaluation criteria. Finally, we introduce various dexterous grasping and manipulation methods, including analytical approaches, machine-learning techniques, and sampling-based methods.