本项目基于sEMG（表面肌电信号）和FES（功能性电刺激）对上肢康复机器人自适应主动控制展开研究。以实现患者主动运动意图个性化建模和康复机器人自适应人机交互为目标，深入探讨基于sEMG的主动康复训练控制方法、康复机器人与FES联合助力控制方法、康复机器人在线康复评价与个性化康复训练策略等方面的关键科学问题与技术，针对上臂和前臂sEMG信号的不同特点进行分别处理，基于模式识别、人工CPG（中央模式发生器）等技术，提出sEMG人体运动意图识别模型的在线学习的新理论和新方法；针对人体上肢复杂的肌肉结构和电生理特性，结合肌骨系统生物力学模型，通过模糊迭代学习方法实现FES与康复机器人的联合助力闭环控制；基于sEMG等多传感器信息融合，探讨康复机器人在线康复评价方法，提出面向临床的个性化康复训练策略。结合临床试验研究形成较为完善的上肢康复机器人自适应主动控制理论和方法，并最终期望达到实用水平。

This project mainly focuses on adaptive and active training methods for upper-limb rehabilitation robot based on surface electromyogram (sEMG) and functional electrical stimulation (FES). With the goal of personalized active motion intention modeling and adaptive human-robot interaction, this project makes an in-depth exploration of active training control method, joint assistance method for rehabilitation robot and FES, and online assessment and personalized training protocols. A novel motion intention modeling method is proposed based on pattern recognition and artificial CPG (Central Pattern Generator) techniques. Joint assistance of rehabilitation robot and FES is implemented based on human musculoskeletal model and iterative learning control method. This project also explores the online rehabilitation assessment method using multi-sensor fusion technique, which facilitates the development of personalized training protocols. All these methods will be evaluated in clinical trials, and expected to achieve the desired result.