在手功能康复中，急切需要训练与辅助功能一体化、手体柔性轻巧、可穿戴便携，且易加工、低成本、安全可靠的手功能康复机器人。本项目将结合人手结构和关节运动特点，利用柔性铰链仿生手指关节，线传动仿生手指肌肉、肌腱运动，形成一种新型线驱动柔性外骨骼手功能康复机器人系统；并通过分析柔性外骨骼运动状态、线槽摩擦力及线变形对末端驱动力的影响，建立手指运动状态下人机耦合瞬时静力学与实时动力学模型，研究利用李群李代数理论简化动力学模型并给出动力学方程高效解法，提出多目标动态优化设计方法；在此基础上建立基于递归神经网络的自适应人机混合协调控制系统，通过参考模型自我学习来满足不同肌力患者的主动运动需求，达到训练与辅助功能统一，满足上述康复需求；阐明人机混合协调控制机制与应用方法。本项目研究成果不仅推动我国柔性外骨骼康复机器人技术进步，并且对人机混合系统的控制策略和人机协同等共性问题的研究具有科学意义和学术价值。

There are some urgent requirements for hand rehabilitation that a robot can implement both functions of training and assist for patients and the robot can be flexible, lightweight, wearable, portable, easy to process, low cost, safe and reliable. This project will propose a novel cable-driven compliant exoskeleton robot for hand rehabilitation that will be designed with flexure hinge as hand joint and cable driven mechanism as hand muscle and tendon by combined the characteristics of hand structure and motion. The relationship between the driving power and the deformation of flexure hinges, the force of friction, and the deflection of the cable. The instantaneous static model and real-time dynamic model during the finger movement will be deduced based on the aforementioned relationship. The dynamic model will be solved by Lie groups and Lie algebras. A dynamical optimal design method for this proposed compliant exoskeleton mechanism will be proposed according to the dynamic model. Meanwhile, a novel self-adaption coordinated control system with man-machine coupling will be deeply researched based on RNN. This control system can meet the different requirements from different patients by means of self-learning from a reference model. Therefore, this kind of rehabilitation robot can promote the patients' enthusiasm for active rehabilitation training, be safe and provide both functions of active and passive motions. In addition, the control theory and its application method will be described clearly in order to boost development of compliant exoskeleton robotic in China. It also has important scientific and academic value on the research of the common problems such as control strategies and man-machine coordination problems of the man-machine hybrid systems.