A Basic Study on Development of Intelligent Motorized Prosthetics

智能电动假肢研制的基础研究

• 批准号: 10838004
• 负责人: YOKOI Hiroshi
• 金额: $ 2.05万
• 依托单位: HOKKAIDO UNIVERSITY
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for Scientific Research (C)
• 财政年份: 1998
• 资助国家: 日本
• 起止时间: 1998 至 2000
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-10838004/
• 关键词: Adaptive Learning Theory; Processing Dura Mater Signals; Adaptability for Individuals; Achieve lost Motor Function; Multi-site Electrode; FPGA; Electrophysiology; Control of Motorized Prosthetics; 脊髄損傷; 電動義装具; 知的制御系; 硬膜外電位; 刺激反応系モデル

This research project is focused on the technical back-up for the people with spinal cord injuries to achieve the motor function, which was lost by an accident. The aim of this research is to develop a methodology on the adaptive control of the motorized prosthetics for individuals using a current LSI technology and an adaptive learning theory in order to measure and process the action potential signal of the dura mater of the spinal cord at the brain-side from the damaged area. The important components of this research are development of an adaptive classification method, reconstruction of a control function using FPGA (Filed Programmable Gate Array), classification of electrophysiological signals of a rat cranial dura mater for a new communication and control, and development of multi-site electrode using micromachining technology.The findings achieved by this research are as follows.1. We applied the adaptive learning theory to classify the human electromyogram as the signal includi … More ng the differences of individuals, and performed to control the prosthetic hands. As a result, the adaptation for individuals and stable classification was verified, and the adaptive learning theory was effective.2. We measured from cranial dura mater, which is closer to the sensory cortex, in an anaesthetized rat electrophysiologically.. The cluster corresponding to the stimuli was generated by the off-line time series analysis using the adaptive learning theory.3. With the same method, we measured the signal from the area, which is concerned with the control of the lower limb, above the Layer-I of the motor cortex using electrodes located at cranial bone. And the result showed the possibility of classification and extraction for the signal corresponding to the movement of lower limb.4. We developed multi-layer, multi-site implantation microelectrode. And the measurement of the signal from the dura mater of the spinal cord with the various implantation methods using wire-electrode or multi-site electrode is tried.5. The signal processing system executable to synthesize an intended mapping function and the experience system by using FPGA was developed. The experience showed that this system has the computability to synthesize the signal processing circuit on the learning theory and the capability of the effective high speed processing in terms of parallel processing.Conclude this research : The aim of this research is to develop the device to classify the signal for individual's characteristics of action potential as the real-time adaptive processing system. For this purpose, we performed to construct adaptive classification method for nerve signal, and showed the possibility of synthesis of the signal processing circuit on the learning theory by FPGA for real-time processing. With considering the result of animal experience, we serve as a stepping-stone to develop the intelligent motorized prosthetics by adaptive classification of signal on the dura mater of the spinal cord. And we made the result of these researches public for dissertation and so on. Less

该研究项目的重点是为脊髓损伤患者提供技术支持，以实现因事故而丧失的运动功能。本研究的目的是开发一种机动假肢的自适应控制方法。个人使用当前的LSI技术和自适应学习理论来测量和处理来自受损区域的大脑侧脊髓硬脑膜的动作电位信号。这项研究的重要组成部分是自适应学习理论的开发。分类方法，使用FPGA重建控制功能（场可编程门阵列），对大鼠颅脑硬脑膜的电生理信号进行分类以实现新的通信和控制，以及使用微加工技术开发多位点电极。本研究取得的成果如下： 1．自适应学习理论将人体肌电图作为包含个体差异的信号进行分类，并进行控制假手，验证了个体的适应性和稳定的分类，建立了自适应学习理论。 2.我们对麻醉大鼠的头颅硬脑膜进行电生理测量，该硬脑膜更接近感觉皮层。利用自适应学习理论，通过离线时间序列分析生成与刺激相对应的簇。3．用同样的方法，我们利用位于颅骨的电极测量了运动皮层第一层以上与下肢控制有关的区域的信号，结果表明了分类和提取的可能性。为4.我们开发了多层、多部位植入微电极，并利用线电极或多部位的各种植入方法测量脊髓硬膜的信号。 5.开发了利用FPGA合成预期映射函数的信号处理系统和体验系统。经验表明，该系统具有基于学习理论的合成信号处理电路的可计算性和能力。有效高速结论本研究：本研究的目的是开发一种针对个体动作电位特征对信号进行分类的装置作为实时自适应处理系统。为此，我们构建了自适应分类。神经信号的方法，并展示了通过FPGA在学习理论上综合信号处理电路进行实时处理的可能性，并考虑到动物经验的结果，作为开发智能电动假肢的垫脚石。硬脑膜信号的自适应分类我们很少将这些研究结果公开用于论文等。

项目成果

Takahiro Kawashima, Jun Hakura, Hiroshi Yokoi, Masatoshi Takita, Yukinori Kakazu: "Identification of Reactive Signal in Rat Frantal Cortex Based on Cellular Field Model" Intelligent Engineering Systems Through Artificial Neural Networks. 643-648 (1998)

Takahiro Kawashima、Jun Hakura、Hiroshi Yokoi、Masatoshi Takita、Yukinori Kakazu：“基于细胞场模型的大鼠皮质反应信号识别”通过人工神经网络的智能工程系统。

Hiroki Yamaguchi, Wenwei Yu, Masaharu Maruishi, Hiroshi Yokoi, Yukio Mano, Yukinori Kakazu: "EMG Automatic Switch for FES Control for Hemiplegics Using Artificial Neural Network"Proceedings of 6th International Conference on Intelligent Autonomous Systems

Hiroki Yamaguchi、Wenwei Yu、Masaharu Maruishi、Hiroshi Yokoi、Yukio Mano、Yukinori Kakazu：“利用人工神经网络对偏瘫患者进行 FES 控制的 EMG 自动开关”第六届智能自治系统国际会议论文集

Tomokazu Shindo, Hiroshi Yokoi, Yukinori Kakazu: "Adaptive Logic Circuits based on Net-List Evolution"Journal of Robotics and Mechatronics. Vol.2, No.2. 144-149 (2000)

Tomokazu Shindo、Hiroshi Yokoi、Yukinori Kakazu：“基于网表进化的自适应逻辑电路”机器人与机电一体化杂志。

Tomokazu Shindo, Kensuke Takita, Hiroshi Yokoi, Yukinori Kakazu: "A study for Environmental Adaptation of Robot Using Evolvable Hardware"Intelligent Engineering Systems Through Artifical Neural Networks. vol.9. 445-450 (1999)

Tomokazu Shindo、Kensuke Takita、Hiroshi Yokoi、Yukinori Kakazu：“使用可进化硬件的机器人环境适应研究”通过人工神经网络的智能工程系统。

H.Yamaguchi,W.Yu,M.Maruishi,H.Yokoi,Y.Mano,Y.Kakazu: "EMG Automatic Switch for FES Control for Hemiplegics Using Artificial Neural Network"Proceedings of 6th International Conference on Intelligent Autonomous Systems. 339-346 (2000)

H.Yamaguchi、W.Yu、M.Maruishi、H.Yokoi、Y.Mano、Y.Kakazu：“使用人工神经网络对偏瘫患者进行 FES 控制的 EMG 自动开关”第六届国际智能自治系统会议论文集。