Abstract—How to design a robotic hand reflecting human hand motion information as much as possible is a constantly exploring problem. In this paper, we propose an approach to mechanical design of compliant underactuated finger for prosthetic hand based on the decomposition of human hand movements. Hand movements are decomposed into primary and secondary motion in PCA coordinate system. The primary motion is achieved in free motion via actuators, and the secondary motion is implemented with mechanical compliance matching statistics characteristic of human motion data. Although analysis and design of single finger is always throughout this paper, the same method can be generalized to the whole hand design and the parameters design of other mechanical configuration. I. INTRODUCTION One of the challenging problems in robotics is implement- tation of human-like movements in unstructured human environment. It is difficult to replicate the function of human hand, because of the complicated anatomical structure and mysterious neuromuscular control system. One approach to reduce this complexity is through synergy, which is from neuroscience and shows that a continuous subspace of configuration space can be used to approximate everyday human hand tasks. Santello et al. [1] investigated grasping poses from mime grasps for a large set of familiar objects via principal components analysis (PCA) and revealed that more than 80% of hand posture information is contained in the first two principal components, which shows that the grasping posture can be expressed as a much lower-dimensional subspace of the hand joint space and reflects significant joint coupling and inter-finger coordination. A similar rule about hand motion during grasping was discovered in subsequent research [2]. Human hand synergy provides a natural modeling paradigm for robotics. Ciocarlie and Allen [3] used the idea to exploit the dimensionality reduction in problems of automated grasp synthesis, and has been applied effectively to derive pre-grasp shapes for a number of complex robotic hands. Brown and Asada [4] designed a mechanical hand in which more or less accurate actuators are connected to different groups of mechanically interconnected joints, with a priority *Resrach supported by the National Basic Research Program of China (973 Program) (Grant No. 2011CB013301), the National Science Fund for Distinguished Young Scholars of China (Grant No. 51025518), and the State Key Program of National Natural Science of China (Grant No. 51335004). Wenrui Chen, Caihua Xiong, Mingjin Liu, and Liu Mao are with School of Mechanical Science and Engineering, Huazhong University of Science and Technology, Wuhan, 432700, China (e-mail: {chenwenrui, chxiong, mjliu, maoliu}@hust.edu.cn). inspired by resemblance to postural synergies observed in human hands. Brown and Asada [4] designed the mechanical hand to replicate the lower-order synergies and ignore higher-order principal components. Introducing a model of compliance in rigid-body system to solve force indeterminacies, Bicchi et al. [5, 6] did not give the method to design parameters but showed quality of grasp is quite robust with respect to parameter values. The synergy approach to the mechanical design of anthropopathic hand can greatly reduce the number of actuators and simplify the control strategy, but it also brings new issues. The postural synergies, i.e. the first few principal components, account for vast majority of hand posture information. However, even though they are small, higher-order principal components do not represent random variability but instead provide additional information about the object [2]. How to design anthropopathic hands to retain high-order information while implementing the principal motion is a difficult problem. In this paper, the approach to embed human hand movement information into robotic hand mechanism is studied. In contrast to the grasp measure of hand such as force distribution, force-closure [5] and robust [7], we are interested in the motion of robotic hand imitating human hand. In particular, we examine the motion characteristics of index finger and implement it in mechanic finger. In different grasp patterns (such as power, precision, lateral, etc.), there are significant independence between and among each finger, and the single finger is allowed visual representation and detailed analysis, which is difficult for the whole hand with more than twenty DOFs [8] [9]. In the work presented in this paper, we extract human finger movement characteristics with PCA and analyze the finger movement behavior in PCA coordinate system, which is emulated using tendon-pulley mechanism. Then, compliance substituted for higher-order principal motions, we propose a quantitative method of mechanical implementing anthropomorphic posture synergy and compliance, and accordingly present a novel anthropopathic fingers which can be embedded in prosthetic hands. II. HUMAN FINGER MOVEMENTS We hope to provide methodological guidance for prosthetic hand design through analysis of human hand motion data. Therefore, the three-joints finger, easy to visualized analysis, will be analyzed as an example to elicit our design ideas. A. Acquisition of human movement data The human hand permits an infinite number of different trajectories to move the fingers from one location in space to Characteristics Analysis and Mechanical Implementation of Human Finger Movements Wenrui Chen, Caihua Xiong, Mingjin Liu, and Liu Mao 2014 IEEE International Conference on Robotics & Automation (ICRA) Hong Kong Convention and Exhibition Center May 31 - June 7, 2014. Hong Kong, China 978-1-4799-3684-7/14/$31.00 ©2014 IEEE 403