【作者】 钱少明；

【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， 2012， 博士

【摘要】 多指机械手作为机器人与外界环境互相作用的末端执行部件一直受到研究人员的关注。传统的机械手爪存在抓持动作单一、自由度少及通用性差等缺点，严重制约了机器人的发展与应用。在果实生产中，收获采摘作业约占整个作业量的40％。随着人口的老龄化和农业劳动力的减少，农业生产成本也将随之提高。因此，发展机械化收获技术，研究开发果实采摘机器人，具有重要的现实意义。本文在以往研究的基础上，进一步深入研究了基于FPA的弯曲关节的基本特性，分析了三自由度手指的输出力特性、提出了新型气动柔性多指采摘机械手的设计目标、机械结构和传感检测单元配置、控制算法，通过模块化设计，能够较好地适应不同种类果实的采摘。主要完成的研究工作如下：（1）提出了由FPA直接驱动的弯曲关节。从静力学的角度，建立弯曲关节的转角及输出力矩静态模型；以热力学第一定律为理论依据，结合弯曲关节的动力学方程，推导了关节的动态模型。对弯曲关节进行了静态特性测试实验，实验结果与仿真结果基本一致；关节转角开环阶跃响应时间大约是1s，关节输出力的开环阶跃响应时间大约是0.5s；采用前馈补偿及PI反馈控制可使关节转角实现精确控制。（2）提出了基于专家控制器的柔性手指指端抓持模型。基于气动柔性弯曲关节设计了一种三自由度手指，采用D-H法建立了三自由度手指的运动学方程及力雅可比矩阵；在分析手指受力状况的基础上，建立了指端输出力与各个关节FPA内腔气压值之间的映射关系，实验结果与指端输出力模型基本吻合。建立了气动柔性弯曲关节的刚度模型，进行了实验研究，其结果与仿真曲线基本一致；进一步分析了三自由度手指指端刚度，并进行了仿真研究。介绍了一种基于关节柔性的手指指端抓持模型，针对不同类型的干扰力，分析了手指指端抓持的稳定性，提出了基于专家控制器的柔性指端抓持控制，并完成了相应的实验研究，取得了良好的控制效果。（3）提出了一种手指指节正压力可控的包络抓持模型。分析了目标物体受力状况，按照手指指节与目标物体之间的各个接触力大小尽量均匀的原则，对目标物体受力进行了优化。建立了关于手指指节接触点所受到的正压力及摩擦力与关节FPA输出力之间的力学模型。使用两个触力传感器，应用杠杆原理建立了接触点正压力及其作用点的测量模型。提出了指节接触点正压力的双闭环控制策略，设置了补偿器对摩擦力进行实时补偿，对压力反馈信号进行微分处理，用以消除压力检测信号中所包含的高频噪声。搭建了实验平台，实验结果表明：手指正压力动态响应时间为约1s，误差稳定在±0.5N范围。（4）分别测定了黄瓜和苹果的抗压特性、果实表面与硅胶表面之间的摩擦系数和果柄切断阻力等特性。（5）详细阐述了多指采摘机械手的设计目标、机械结构、传感单元、控制单元和切割器。提出了两类不同精度等级的多指采摘机械手，分别为标定多指采摘机械手和实用多指采摘机械手。标定多指采摘机械手配备了关节位置传感器、多维指端力传感器、指节触力传感器及压力比例阀；实际采摘作业时使用实用多指采摘机械手，它只对各个关节FPA中的压力值进行闭环反馈控制。用力学分析的方法建立了黄瓜和苹果的抓持模型，分析了气动驱动器中的气压值与抓持能力之间的关系。研制了可用于黄瓜采摘的机械手，采摘效果良好，黄瓜抓持成功率为90%，黄瓜果柄割断成功率为100%，采摘时间约为3s。本文研究的新型气动采摘多指机械手，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】 As executing manipulator for robot to interact with outside environment, themulti-fingered robot hands have been paid much attention. Earlier conventional mechanicalhand is of simple structure, fewer degrees of freedom DOFs, poor adaptability and othershortcomings, which restricts the development and application of robots.The work of picking fruit mainly depends on manpower. The picking work is a mosttime-consuming and laborious production process. In order to improve the efficiency ofpicking fruit the mechanization should be attached importance to. With farms population isobviously reduced and aging population has been gradually become a common trend in theworld, development of fruit picking robot is of enormous economic potential and broadmarket prospects. Found on the previous research, bending joint driven by flexible pneumaticactuator FPA is studied profoundly in this paper. The3-DOF finger based on FPA and themulti-fingered hand for fruit picking are proposed. The main research work in this paper is asfollows:(1) Bending joint driven by FPA directly is proposed. Based on statics and elasticity, thestatic model about angle and output force of the bending joint is obtained. According to thefirst law of thermodynamics, combined with the joint’s dynamic equation, the dynamic modelof the bending joint is established. Experiments are carried out to verify the static model. It isconcluded from theoretical and experimental results that the experimental curve matches withthe simulated curve. Experimental results show that the angle step response time of the joint isabout1s, and the force step response time of the joint is about0.5s. The accurate angle can begot based on PI feedback control and feed forward compensation.(2) Finger-tip grasping model based on3-DOF fingers is proposed. A3-DOF fingerbased on bending joint is design. Using D-H method, kinematics equation and Jacobin matrixof the finger are educed. The statics of the finger is analyzed, and then static model aboutfingertip force and the values of the pressured air in the three FPAs is built. Experiments arecarried out to verify the static model, and it is concluded from theoretical and experimentalresults that the experimental force value matches with the simulated force value basically. Stiffness model about the bending joint is established. Experiments of the stiffness about thebending joint are carried out, and the theoretical and experimental results show that theexperimental curve matches with the simulated curve. Stiffness about the fingertip is alsoanalyzed. This paper presents a new flexible fingertip grasping mode. In view of differenttype of disturbing forces, stability of the fingertip grasping is analyzed, and the fingertip graspbased on expert controller is proposed. Experiment was done to verify the control algorithm.(3) Envelop grasping model based on3-DOF fingers is proposed. Based on the principleof equalization, the interaction forces between the object and knuckle is optimized. The staticmodel about the values of the pressured air in FPAs and the positive pressure on knuckle orcorrelative friction force is built. The measure model about the positive pressure and itsposition, based on leverage principle depending on two touch-force sensors, is obtained.Using a series dual loops control method and the compensator about friction force, thepositive pressure can be controlled precisely. In order to filtrate the high-frequency noise, thefeedback signal of the pressured air in FPA is processed by differential analysis. Theexperimental results show that dynamic response time of the positive pressure is about1s andthe steady-state deviation is less than±0.5N.(4) The compressibility characteristic of cucumber and apple, the cutting characteristic ofcucumber peduncle and apple peduncle, the friction coefficient between cucumber coat andsilica gel, the friction coefficient between apple coat and silica gel have been testedrespectively.(5) Design objective, mechanical structure, sensor system and control system ofmulti-fingered robot hand for picking are illustrated in detail. The calibration multi-fingeredrobot hand and the practicable multi-fingered robot hand are proposed, which have differentprecision ratings. The calibration multi-fingered robot hand is equipped with joint anglesensors,5-component force/torque sensor, touch force sensors and proportional pressure valve,and the practicable multi-fingered robot hand is only equipped with proportional pressurevalve which can be used in loop control on the value of the pressured air in FPA. The holdingmodels of cucumber and apple are established with mechanical analysis method, therelationships between the pressure value of compressed air in the FPA and picking capacityare analyzed. A picking robot hand for cucumber is developed. The effect of cucumberpicking using this hand is well, and the success ratio of picking cucumber is90percent. Thesuccess ratio of cutting cucumber peduncle is100percent, and the time of picking a cucumberis about3s.A new type of multi-fingered robot hand for picking, which is directly driven by FPA developed by our research team, is proposed in this paper. This robot hand has characteristicsof simple structure, easy control, and easy miniaturization and so on. It has better passiveflexibility, and without lack of stiffness. It is suitable for application situation that requiresrelatively high flexibility and low response speed of the fingers, such asagricultural harvesting robot, finger rehabilitation robot and so on.更多还原