Improvement of Control Performances of Rubber Artificial Muscle Manipulator and Its Application for Human-Robot Collaboration Task

橡胶人工肌肉由于其柔韧性的基本特征，适合于人类的协作任务。然而，由于其阻尼性不足，要实现令人满意的操纵性能并不容易。然后在机械手关节上安装一种阻尼力可调的电流变阻尼器(ER阻尼器)，采用气动伺服主动控制和电流变阻尼器阻尼器相结合的控制方法，获得了满意的控制效果。同时，将橡胶人工肌肉机械手作为康复机器人应用于手臂功能恢复训练。实验结果表明，本文提出的基于阻抗控制的多种功能恢复训练模式统一实现方案是有效的。为了实现自主式康复机器人，提出了一种基于二阶机械阻抗模型的手臂特性辨识方法。在实施训练操作的同时，使用自适应识别器识别阻抗模型的每个参数。因此，可以同时估计训练操作中的恢复程度，这有望成为实现自主康复机器人的一种有效方法。以下两个问题是有待进一步研究的问题：一是如何将识别的参数反映在功能恢复训练操作上；二是开发人机协作系统，以实现训练操作与识别操作的统一。

A rubber artificial muscle is suitable for the human collaborating task owing to its essential features of flexibility. However, it is not easy to implement a satisfactory manipulation performance since its damping characteristic is not sufficient. Then an electric rheological damper (ER damper) whose damping characteristic can be regulated is attached to the manipulator's joint, which leads to a satisfactory control performances by applying an active control of pneumatic servo with a damping control using an ER damper.At the same time, the rubber artificial muscle manipulator is applied to a function recovery training of human arms as a rehabilitation robot. Experiments results showed the effectiveness of a proposed unified implementation scheme for several function recovery training modes by using impedance control method. Also an identification method of characteristic of rehabilitant's arm was proposed for the purpose of a realization of an autonomous rehabilitation robot, in which characteristic of arm is estimated by the second order mechanical impedance model. Each parameter of the impedance model is identified using adaptive identifier at the same time with an implementation of a training manipulation. Consequently the recovery level can be estimated during the training manipulation simulataneously, which is expected to be an effective method to realize an autonomous rehabilitation robot.Following two matters are left as a future work, one is how the identified parameters should be reflected on the function recovery training manipulation and the other is the development of constitution of human-robot collaboration system in order to unify the training manipulation with an identification one.