【作者】 卫玉芬；

【导师】 李小宁；

【作者基本信息】 南京理工大学 ， 机械制造及其自动化， 2006， 博士

【摘要】 气动肌肉驱动的机器人手臂是一类具有良好柔顺性能的机器人手臂，在很多应用领域具有突出的优势。在气动肌肉驱动的机器人手臂系统中，关节刚度调节、位置控制及手爪的柔顺夹持是其中最为重要的关键技术问题。本文以气动肌肉驱动的机器人手臂(含手爪)系统为背景，对以上的关键技术进行了系统的研究，并设计了一套实验系统对研究结果进行了详细的验证。本文的主要工作如下： 在机器人手臂系统设计方面，进行了详细的方案设计和系统软硬件调试，确定了手腕俯仰关节和手指关节分别采用两种不同规格气动肌肉进行驱动的方案，该系统由前臂、手腕、手爪三个部分组成。 在关节刚度调节方面，重点对气动肌肉的驱动模型和气动肌肉驱动关节的刚度特性进行深入研究。首先分析了气动肌肉的基本模型，在理论分析的基础上，提出利用收缩率参量修正来补偿多种误差的方法，并给出了气动肌肉的改进模型；然后利用气动肌肉的改进模型对关节的输出转矩进行了详细推导；最后给出关节刚度的详细表达式，并提出了一种利用简单、有效的比例刚度调节方法。实验结果验证了气动肌肉的改进模型和上述关节刚度调节方法的合理性、有效性。 在关节位置控制方面，进行了如下几个方面的工作：首先理论分析和实验证明了PID控制器的可行性和有效性，但关节的响应时间太慢不能满足整个机器人轨迹规划的需要，因此提出采用具有自适应、自学习功能的单神经元自适应PSD控制器来改善关节的控制性能，该控制器可以同时实现对增益、比例、积分和微分四个参量的自适应调整，实验证明该控制器使气动肌肉驱动的关节具有响应时间短、位置精度高、鲁棒性好等显著优点。 在手爪的柔顺夹持控制方面，进行了以下几个方面的工作：设计了由四根气动肌肉驱动的仿人两指手爪结构；建立了气动肌肉驱动手指夹持力的理论模型；分析了肌肉初始输入压力和初始收缩率对手指张开范围和夹持力的影响。分析结果表明，手指内外侧肌肉的输入压力是决定手指夹持力的重要因素。此外，还对手指运动过程的动静态特性和夹持力进行了系统的测试。实验表明，该气动肌肉驱动的手爪可以实现对多类物品的柔顺抓取。更多还原

【Abstract】 The robot arm actuated by pneumatic muscles is one kind of the robot arm with good compliance and it is superior to other robot arms in some cases. For the robot arm system actuated by pneumatic muscles, the joint stiffness regulation, the joint position control, and the finger grasping force control are three key techniques. In this dissertation, the three techniques are studied and some effective methods are proposed. Moreover, a robot arm prototype system actuated by pneumatic muscles is designed to verify those proposed methods. Main works of this dissertation are as follows:The scheme of the robot arm prototype system is presented and this system is composed of three subsystems, i.e. forearm, wrist and gripper. Moreover, the detailed implemented scheme is proposed and the prototype system, which includes the software and hardware, are debugged.For the joint stiffness regulation, the static drive model of pneumatic muscle is improved and stiffness characteristic of the joint is analyzed. The detailed work includes:An improved static drive model of pneumatic muscle is proposed, it can compensate the complicated errors due to various factors by adjusting the contraction ratio. On the basis of the model, the static drive model of the joint is introduced. An indirect measurement method of the joint stiffness is proposed by measuring the static parameters of the joint, and its detailed equation of the joint stiffness is given. Some experiments are performed to validate these works and the results show their effectiveness.For the joint position control, the detailed work includes:Theoretical analysis and experiments proved that PID controller for the joint position control is feasible, effective, However its slow response time doesn’t satisfy the need of the trajectory planning for the 4 DOF articulated robot, and so the author proposed to adopt the adaptive and self-learning single neuron PSD controller to improve motion performance of the joint, the controller can adjust adaptively all four parameters, i.e. proportion, integral, differential and gain simultaneously. Experiments show that a good joint position accuracy and a short step response time are achieved.For the finger grasping force control, the detailed work includes:The structure of humanoid two-fingers gripper is designed. The theoretical model of the finger grasping force is developed. The effect of the initial input pressure and the initial更多还原