实现下肢假肢智能仿生控制的神经功能重建及行走意图识别方法

Walking is one of the basic functions of human beings to independently live and normally participate in social activities and thus the recovery of walking function after lower-limb amputations would be significantly meaningful. The currently available leg prosthetic systems can not satisfy yet the expectations and demands of lower-limb amputees due to their low recognition accuracy of ambulation intentions and lack of gait control stability. With an attempt to solve these problems of the current lower-limb prostheses, this project will conduct the fundamental research works on three major and important aspects related to the control of a lower-limb prosthesis, accurate recognition of ambulation intentions, intelligent control of prosthetic gait, and design of high-performance prosthetic joints. Considering that the residual muscles after leg amputations usually not provide sufficient electromyography (EMG) information for ambulation-intention recognition, a novel method of neural function reconstruction, so-called hybrid reinnervation of targeted nerves & muscles (HR-TNM), will be used in the project to reconstruct the neuromuscular functions of an amputated leg, which could provide physiologically appropriate myoelectric signals during ambulation. By combining the reinnerated EMG signals from the residual limbs with the biomechanical signals from the interface between amputated leg and prosthesis and extracting the features of these signals, some advanced pattern-recognition algorithms will be developed to recognize the walking intentions accurately and automatically. The intelligent control strategy for lower-limb prostheses will be exploited based on the walking intentions and intelligent algorithms. By embedding the intelligent control strategy into a leg prosthesis that will be assembled with high-performance prosthesis joints, the stable and natural ambulation on different topographic conditions will be achieved by wearing the leg prosthesis. And a continuous and natural transition from one ambulation mode to another will be also done. The successful implementation of this project will be able to fill the blankness of the fundamental theories and technologies in neuromuscular function reinnervation and the intelligent control of a leg prosthesis in China.

行走是人类独立生活和正常参与社会活动的基本功能之一，下肢截肢后行走功能的恢复具有非常重要的意义。由于行走意图识别精度低、步态控制稳定性差等问题，目前的下肢假肢系统还不能满足截肢者的祈求。针对下肢假肢控制存在的问题，本项目拟重点开展行走意图精确识别、假肢步态智能控制及高性能假肢关节设计等方面的研究。针对残肢肌肉不能为行走意图识别提供足够肌电信息的问题，利用称之为靶向神经-肌肉重建术（HR-TNM）的神经功能重建创新方法，为残肢实施神经功能重建，获取更多神经肌肉电信号；将重建的残肢肌电信号与生物力学信号融合，提取对应行走模式的特征参数，研发先进的模式识别算法，实现行走意图的精确识别；依据行走意图研发下肢假肢的智能控制策略，并结合高性能假肢关节实现假肢在不同地形条件下的稳定行走及步态模式间的无缝和自然切换。本项目的实施有望填补我国在下肢神经功能重建及智能控制理论与技术方面的空白。

行走是人类独立生活和正常参与社会活动的基本功能之一，下肢截肢后行走功能的恢复具有非常重要的意义。由于行走意图识别精度低、步态控制稳定性差等问题，目前的下肢假肢系统还不能满足截肢者的祈求。针对下肢假肢系统存在的问题，本项目重点开展了下肢假肢仿生控制的“行走意图自动精确识别”、“假肢步态智能控制策略”及“高性能智能假肢关节设计”等方面的研究。主要的研究内容及成果包括：（1）针对截肢后肢体肌电信号源不足，首先综合两种神经功能重建技术（TMR和TNFR）两种方法的优点，提出了靶向神经-肌肉重建术（HR-TNM），开展了神经肌肉功能重建的动物和人体模型研究。研究结果表明：新提出的方法解决了TNFR术式中神经端面可能无法完全吻合的问题，又充分利用了靶向肌内原有的神经网络形态；然后利用聚乙二醇（PEG）与功能电刺激（FES）干预重建后的靶向神经-肌肉，探明使用功能电刺激（FES）对HR-TNM模型进行干预可以增强神经再生与改善肌肉废用性萎缩，使得术后恢复时间降低了50%，提高HR-TNM的重建效果；（2）研制了新型的柔性可拉伸传感器以及复合信号采集系统，实现了生物力学生物电学等多模信号的高精度同步采集，并将不同组合的多模信号应用于行走意图识别、步态参数识别以及环境识别；同时优化了传感器和运动意图识别算法的各种参数，识别准确率达到90%以上；然后依据假肢使用者的行走意图、步态参数和环境识别参数，利用有限状态机实现了步态以及状态的自动转换，并使用阻抗控制方法实现了假肢的变刚度控制，达到了假肢仿生控制的目的；（3）在运动意图识别和假肢步态控制方法研究的基础上，开发了一款综合性下肢假肢平台，包括了髋关节模拟平台、轻量的被动踝关节、动力踝关节和动力膝关节假肢，重点研究了三种不同动力型膝关节假肢，最后开发了一款基于一体化星行减速电机的膝关节和被动踝关节假肢，招募肢体健全者和下肢截肢者进行假肢穿戴行走测试，实现了假肢在不同地形下的实时、稳定、自然的行走。