【作者】 王志恒；

【导师】 张立彬； 杨庆华；

【作者基本信息】 浙江工业大学 ， 机械电子工程， 2011， 博士

【摘要】 末端执行器是机器人与环境发生交互作用的最重要部件。传统的末端执行器存在结构简单、自由度少及适应性差等缺点,严重影响了机器人的发展与应用。机器人多指灵巧手是一种模拟人手结构、功能及尺寸,具有多自由度、多关节的新型末端执行器,具有良好的灵活性和适应性,为机器人的应用起到了积极推动作用,具有很好的研究意义与应用价值。20世纪90年代后,一般功能的工业机器人的应用趋于饱和,机器人的应用逐渐延伸至农业、医疗康复、服务娱乐等领域中,其抓取目标往往是一些脆、嫩或有生命体的对象,因此,要求机器人多指灵巧手具有很好的柔性。本文总结了现有灵巧手存在柔性差、结构复杂、难以控制等缺点,并在此基础上提出了一种新型的气动驱动多指灵巧手,命名ZJUT多指灵巧手,其具有良好的被动柔性,同时在某些方面弥补了现有灵巧手的不足。主要完成的研究工作如下:(1)提出了采用FPA直接驱动,模拟人手指侧摆运动的侧摆关节。从静力学角度,以弹性力学为基础,对驱动元件FPA自由端进行力平衡分析,建立了侧摆关节的静态模型;根据热力学第一定律,结合关节的动力学方程,建立了关节的动态模型;实验研究了侧摆关节的静动态特性,验证了数学模型的正确性,分析了理论与实际曲线误差存在的原因;采用了串联双闭环控制方法,对关节的转角及输出力矩进行了控制研究,结果表明:期望角度为15°时,关节转角闭环动态响应时间约为0.3s,稳态相对偏差小于0.65%;期望输出力矩为188Nmm时,闭环输出力矩动态响应约为0.3s,稳态相对偏差小于1.5%;侧摆关节可控性高,满足多指灵巧手关节设计要求。(2)提出了具有4个关节、4个自由度的拟人手指;采用模块化设计方案,将5个完全相同的手指装配在拟人手掌上,构成ZJUT多指灵巧手的本体结构;采用仿生学的优化方法确定了灵巧手的结构参数及整手的布局方案,通过仿真实验验证了该手的抓持能力;研制了ZJUT灵巧手的传感系统;总结了ZJUT灵巧手的特色之处。(3)采用D-H法建立ZJUT灵巧手手指的运动学方程;分析得出手指的逆运动学存在冗余解问题;提出一种改进的自适应遗传算法IAGA,仿真实验表明:该算法收敛速度快,鲁棒性强,能够有效的解决手指的逆运动学求解问题;对手指进行了位置跟踪实验,实验结果表明:对于不同的目标位置,指尖位置跟踪动态响应时间小于0.5s,位置最大误差为1.12mm。(4)基于微分运动学理论,建立了ZJUT灵巧手手指的静力学模型,完成了手指静力跟踪的半闭环控制实验,验证ZJUT灵巧手手指输出力便于控制的特点;分析了ZJUT灵巧手具有良好被动柔性的特点;基于指尖五维力传感器,提出了一种模糊自适应指尖力动态跟踪控制策略,完成了手指指尖力动态跟踪实验,结果表明:该控制策略能够在未知环境下,实现对手指指尖力快速、精确的动态跟踪,响应时间约为1s,跟踪误差稳定在±0.15N范围内。(5)定义了ZJUT灵巧手手掌最佳抓取姿态;提出了基于ANFIS的抓取模型辨识方法,仿真结果表明:该方法能够很好的对目标物体进行模型辨识,同时具有很好的辨识精度和收敛性;提出了一种目标物体规则化等价方法,大大提高了模型辨识效率;介绍了基于线性约束梯度流抓取力优化方法;提出了一种多指灵巧手通用抓取规划方案;最后,完成了ZJUT灵巧手的抓取规划综合实验,结果表明:ZJUT灵巧手能够对典型形状的未知目标物体进行模型辨识,根据目标物体的模型,选择合理的参与抓取手指数量及抓取点位置,对目标物体实施稳定抓取,并能够完成抓取力的优化。本文研究的新型气动多指灵巧手,采用课题组自主研发的气动柔性驱动器FPA直接驱动,具有结构简单,便于控制,易于小型化等特点,具有良好的被动柔性,同时不缺乏刚度。适合应用在一些柔性要求相对较高,对响应速度要求相对较低的场合,如农业采摘机器人,手指康复机器人等。更多还原

【Abstract】 End-effector is an important manipulator for the robot to interact with environment. Because of simple structure, fewer degrees of freedom DOFs, poor adaptability and other shortcomings, earlier conventional end-effector is one of the most important factors which restricts the development and application of robots. Simulating the structure, function and size of human hand, robot multi-fingered dexterous hand with multiple joints and multiple DOFs is a new kind of end-effector which has better dexterity and adaptability as well as plays a positive role to promote application of robots. Therefore robot multi-fingered dexterous hand has both great research significance and application value.After 1990s, industrial robots with the general function have become saturated. Application of robots has been extending to agricultural, medical and service areas from industrial area, and its grasping targets are always crisp, tender or living objects. So, it requires the multi-fingered dexterous hand should have great flexibility. Based on the shortcomings of existing dexterous hands summarized in this paper, a new type of air-driven multi-fingered dexterous hand, named ZJUT Hand, is proposed. ZJUT Hand has good passive flexibility and can make up for the deficiency of existing dexterous hands in some ways. The main research work in this paper is as follows:(1) Side-sway joint driven by FPAs directly is proposed which can simulate swing movement of human hand. Based on statics and elasticity, force equilibrium equation of the FPA free end is established. And then the static model of the side-sway joint is obtained. According to the first law of thermodynamics, combined with the joint’s dynamic equation, the dynamic model of the side-sway joint is established. Experiments are carried out to test the static and dynamic characteristics of the joint and verify the mathematical model. There is certain error between experimental curve and theoretical curve, and the error cause is analyzed. Using a series dual loop control method, the output angle and output force of joint are controlled. The results show: when expected output angle is 15°, the loop dynamic response time is about 0.3s, and the steady-state relative deviation is less than 0.65%. When expected output force is 188Nmm, the loop dynamic response time is about 0.3s, and the steady-state relative deviation is less than 1.5%. The side-sway joint can meet the requirement for designing multi-fingered dexterous hand joint. (2) 4-DOF anthropomorphic finger with four joints is proposed. To achieve a high degree of modularity, five identical fingers and a palm constitutes the body structure of ZJUT Hand. The structural parameters of the fingers and the layout program of the hand are determined by the bionic optimization method. Simulation experiments are carried to verify the grasping ability of ZJUT Hand. The multi-sensory system for ZJUT Hand is developed.(3) Using D-H method, direct kinematics equation of ZJUT Hand finger is established. Analytical analysis show that there exits a redundant problem for solving inverse kinematics of the finger. An improved adaptive genetic algorithm IAGA is proposed. Simulation experimental results show that the IAGA can effectively solve inverse kinematics problem of the redundant finger. Experiments of finger position tracking are carried out, and the experimental results show: for different expected fingertip position, dynamic response time of position tracking is less than 0.5s, and the maximum position error is 1.12mm.(4) Based on differential kinematics theory, the static model of the finger is established. Semi-closed loop control experiments for static force tracking of the finger are carried out. It is concluded from experimental results that fingertip output force can be controlled easily. Characteristics of good passive flexibility of ZJUT Hand are summarized. Based on 5-component force/torque sensor intalled on the fingetip, a control strategy for dynamic adaptive fuzzy fingertip force tracking is proposed. Experiments of dynamic fingertip force tracking are completed. Experimental results show that the control strategy can realize fast and accurate dynamic fingertip force tracking under an unknown environment, and the force tracking error is within±0.15N.(5) The optimal grasping posture of ZJUT Hand’s palm is defined. A new method for grasping model identification based ANFIS is proposed. Simulation experimental results show that the method can easily establish the model of the target objects, and has very good recognition accuracy and convergence. Equivalent rules of target objects are presented and greatly improve the efficiency of identification. A general grasp planning program of multi-fingered dexterous hand is proposed. Finally, the grasp planning experiments for ZJUT Hand is completed. Experimental results show that ZJUT Hand is able to construct the models of two typical target objects, choose a reasonable number of fingers for grasping and position of contact points, perform stable grasping of the objects and complete the grasping force optimization.A new type of multi-fingered dexterous hand, ZJUT Hand, proposed in this paper is directly driven by flexible pneumatic actuator FPA developed by our research team. ZJUT Hand has characteristics of simple structure, easy control, and easy miniaturization and so on. It has better passive flexibility, and without lack of stiffness. It is suitable for application situation that requires relatively high flexibility and low response speed of the fingers, such as agricultural harvesting robot, finger rehabilitation robot and so on.更多还原