机器人学基础第5章机器人控制

1、机器人学基础 Fundamentals of Robotics,智能科学基础系列课程国家级教学团队 “机器人学”课程 主讲：蔡自兴 谢斌 中南大学 2010,Fundamentals of Robotics,1,Robot Hubo Made in China Trained in USA,2,Fundamentals of Robotics,HUBO Robot,Hero，Im Hubo. I am a very unique robot. The more we play the more I can do! I can play games, dance and even sing to

2、 you! Lets be good friends!,3,Fundamentals of Robotics,中南大学 蔡自兴，谢 斌 zxcai, 2010,机器人学基础第五章 机器人控制,4,Ch.5 Robot Control,Fundamentals of Robotics,Fundamentals of Robotics,Chapter 5 Robot Control 第五章 机器人控制 5.1 Basic Principles of Robot Control 5.2 Position Control of Robots 5.3 Hybrid Control of Force/Po

3、sition of Robots 5.4 Intelligent Control of Robots 5.5 Summary,Examples Robot HuBo PUMA 560 Watch a video of BigDog,PUMA 560 Precise Universal Machine for Assembly An industrial robot,Robot HuBo A mobile robot Designed in USA Made in China Can play games, dance and sing,Watch vedio BigDog Made by Bo

4、ston Dynamics Co. US. A military robot which has amazing mobility and adaptability. It can climb up 35 slopes, carry more than 40 kg equipment. Impression: Kinem. + Dynam. + ?,10,Chapter 5 Robot Control,5.1 Basic Principles of Robot Control 机器人的基本控制原则 The issue of robot control is closely related to

5、 robotic kinematics and dynamics. (how learn?). From the point view of control, The robotic system is a representative redundant, multi-variable and nonlinear control system in nature, and a complex coupling dynamic system. Every control task it self is a dynamical task.,机器人学基础,11,5.1 Basic Principl

6、es of Robot Control,Classification of Controllers non-servo control servo control position/speed feedback control force (torque) control Hybrid F-P control sensor-based control nonlinear control,5.1.1 Basic Control Principles,5.1 Basic Principles of Robot Control,adaptive control hierarchical contro

7、l optimal control fuzzy control neurocontrol other intelligent controls ,12,5.1.1 Basic Control Principles Industrial robot controller can be divided into single-joint (link) controller and multi-joint (link) controller. Robot control depends on its head, that the development of the processor. Vario

8、us control methods can be used according to the actual working conditions.,5.1 Basic Principles of Robot Control,5.1.1 Basic Control Principles,13,主要控制变量（Control Variables） Control variables of joints of a manipulator,5.1 机器人的基本控制原则,14,控制层次 Control Levels Level 1： Level of AI Level 2 ： Level of cont

9、rol model Level 3 ： Level of servo system,5.1.1 Basic Control Principles,5.1 机器人的基本控制原则,5.1.1 Basic Control Principles,Robot control is the control of bidirectional equation: 末端执行装置的状态是由任务轴的许多参数表示的，它是机器人运动模型矢量X的分量。要控制矢量X 随时间变化的情况，即 ，它表示末端执行装置在空间的实时位置。只有当关节移动时，X 才变化。用矢量 表示关节变量与受控量关系关节变量，即 至 。 各关节具有运动

10、学模型C1至C6，这些模型构成关节矢量 ，并由各传动电动机的力矩矢量 经过变速机送到各个关节。在电流或电压矢量 提供的动力作用和微处理机的控制下，这些电动机产生力矩 。,16,控制层次 Control Levels 1st Grade: Artificial Intelligence level 2nd Grade: Control model level 3rd Grade: Servo system-level,5.1.1 Basic Control Principles,5.1 Basic Principles of Robot Control,5.1.2 Examples of Ser

11、vo-control System 伺服控制系统举例,Servo driven system of hydraulic cylinder,17,5.1 机器人的基本控制原则,Electro-hydraulic servo control system,18,5.1.2 Examples of Servo-control System,5.1 机器人的基本控制原则,19,Manipulator consists of a series of links, whose dynamic characteristics are highly non-linear. We need to build u

12、p the mathematical model to control the motor-driven manipulator. Two assumptions in the designing of model: The manipulator is an ideal rigid body without friction and gap. Only one degree of freedom for each link, either translation or rotation.,5.2 Position Control of Robot 机器人的位置控制,5.2 Position

13、Control of Robot,20,5.2.1 Modeling of D.C Control System 直流传动系统的建模,Transfer function and equivalent diagram,5.2 机器人的位置控制,根据电力传动原理可求得下列传递函数： 电动机的开环传递函数，见式5.16 电枢控制直流电动机的传递函数(1)，见式5.23,22,Speed regulation of DC motor,5.2 机器人的位置控制,5.2.1 Modeling of DC Control System,23,Basic control structures,5.2.2 St

14、ructure of Position Control 位置控制的基本结构,5.2 机器人的位置控制,24,Servo control structure of PUMA,5.2.2 Structure of Position Control,5.2 机器人的位置控制,25,Servo control structure for PUMA robot,5.2.2 Structure of Position Control,5.2 Position Control of Robot,26,Servo control structure for PUMA robot,5.2.2 Structure

15、 of Position Control,5.2 Position Control of Robot,27,5.2.3 Position Controller of Single Joint 单关节位置控制器,Structure of Position control system,5.2 Position Control of Robot,28,Transfer function of a single-joint controller,5.2.3 Position Controller of Single Joint,5.2 Position Control of Robot,29,5.2

16、.3 Position Controller of Single Joint,5.2 Position Control of Robot,30,Determinate Parameters and the Steady-State Error (SSE) Steady-State Error (SSE),5.2.3 Position Controller of Single Joint,5.2 Position Control of Robot,5.2.4 Position Controller with Multi-Joint 多关节位置控制器,31,Lagrangian dynamic e

17、quation,5.2 Position Control of Robot,5.2.4 Position Controller with Multi-Joint,5.2 Position Control of Robot,32,5.2.4 Position Controller with Multi-Joint 多关节位置控制器,33,Compensation of Coupled Inertia 耦合惯量补偿,5.2 Position Control of Robot,34,5.3 Hybrid Position/Force Control of Robots 机器人的力和位置混合控制,5.

18、3.1 Schemes of Hybrid Position/Force Control 力和位置混合控制方案 Active Stiffness Control主动刚性控制 如果希望在某个方向上遇到实际约束，那么这个方向的刚性应当降低，以保证有较低的结构应力；反之，在某些不希望碰到实际约束的方向上，则应加大刚性，这样可使机械手紧紧跟随期望轨迹。于是，就能够通过改变刚性来适应变化的作业要求。,5.3 Hybrid Position/Force Control of Robot,35,5.3.1 Schemes of Hybrid Control,Raibert-Craig Position /

19、Force Hybrid Controller 雷伯特-克雷格位置/力混合控制器,5.3 Hybrid Position/Force Control of Robot,36,Raibert-Craig Position / Force Hybrid Controller,对R-C控制器进行如下改进： 在混合控制器中考虑机械手的动态影响，并对机械手所受重力及哥氏力和向心力进行补偿； 考虑力控制系统的欠阻尼特性，在力控制回路中，加入阻尼反馈，以消弱振荡因素。 引入加速度前馈，以满足作业任务对加速度的要求，也可使速度平滑过渡。 改进后的R-C力/位置混合控制系统结构图如图5.17所示。,5.3 Hy

20、brid Position/Force Control of Robot,37,5.3 Hybrid Position/Force Control of Robot,Raibert-Craig Position / Force Hybrid Controller,38,5.3.1 Schemes of Hybrid Control,操作空间力和位置混合控制系统,5.3 Hybrid Position/Force Control of Robot,5.3.2 Control Rule Synthesis of Hybrid Sys. 力和位置混合控制系统控制规律的综合,位置控制规律 Positi

21、on Control Equation of the system controller: Dynamic Equation of a closed-loop system: Let,39,5.3 Hybrid Position/Force Control of Robot,40,力控制规律 Force Control 令图5.17中的位置适从选择矩阵 S=0，控制末端在基坐标系z0方向上受到反作用力。设约束表面为刚体，末端受力如图5.19所示，那么对三连杆机械手进行力控制时有力控制选择矩阵：,5.3.2 Control Rule Synthesis of Hybrid Sys.,5.3 Hy

22、brid Position/Force Control of Robot,力控制规律 Force Control,Dynamic Equation of a closed-loop system: Results show that on the force control joint 1 does not work, joints 2 and 3 are effective.,41,5.3 Hybrid Position/Force Control of Robot,42,力和位置混合控制规律 Hybrid Control 设约束坐标系与基坐标系重合。如果要求作业在基坐标系的z0方向进行力控

23、制，在某个与x0y0平面平行的约束面上进行位置控制，则适从选择矩阵为 位置： 力：,5.3.2 Control Rule Synthesis of Hybrid Sys.,5.3 Hybrid Position/Force Control of Robot,5.4 Intelligent Control of Robots 机器人的智能控制,5.4.1 Classification of Intelligent Control 智能控制系统的分类 递阶控制(Hierarchical Control) 专家控制(Expert Control) 模糊控制(Fuzzy Control) 学习控制(L

24、earning Control) 神经控制(Neuro Control) 进化控制(Evolution Control),43,5.4 Intelligent Control of Robot,5.4 Intelligent Control of Robots 机器人的智能控制,5.4.1 Classification of Intelligent Control 智能控制系统的分类 递阶控制系统(Hierarchical Control System) 组织级代表控制系统的主导思想，并由人工智能起控制作用。 协调级是上（组织）级和下（执行）级间的接口，承上启下，并由人工智能和运筹学共同作用。

25、 执行级是递阶控制的底层，要求具有较高的精度和较低的智能，它按控制论进行控制，对相关过程执行适当的控制作用。,44,5.4 Intelligent Control of Robot,Hierarchical Control System,Structure of a hierarchical control system The control intelligence is hierarchically distributed according to the principle of increasing precision with decreasing intelligence (IP

26、DI).,45,5.4 Intelligent Control of Robot,46,hierarchical control system of PUMA 600 with vision feedback.,Hierarchically Control System,5.4 Intelligent Control of Robot,47,5.4.1 Classification of Intel. Control,专家控制系统(Expert Control System) Almost all of the expert control system (controller) contai

27、ns the knowledge base(知识库), reasoning engineer (推理机), rule set (控制规则集) and/or control algorithm.,5.4 Intelligent Control of Robot,48,5.4.1 Classification of Intel. Control,模糊控制系统(Fuzzy Control System) A new mechanism of control law of knowledge-based (rule-based) and even language-description. An im

28、proved alternative method to non-linear control.,5.4 Intelligent Control of Robot,49,5.4.1 Classification of Intel. Control,学习控制系统(Learning Control System) Four main functions of learning control: search, recognition, memory and reasoning.,5.4 Intelligent Control of Robot,On-line learning control sy

29、stem,off-line learning control system,50,5.4.1 Classification of Intel. Control,神经控制系统(Neuro-Control System) Control system based on Artificial Neural Network (ANN-based control), abbreviate as neural control or NN control.,5.4 Intelligent Control of Robot,Supervised neural controller structure,51,5

30、.4.1 Classification of Intel. Control,进化控制系统(Evolution Control System) Evolution and feedback are two basic regulatory mechanisms complementary to each other. Combination of the two mechanisms produces a new intelligent control method - evolutionary control. Evolutionary control simulate the evoluti

31、on mechanisms of biosphere, improve the autonomy, creativity and learning ability of the system.,5.4 Intelligent Control of Robot,52,5.4 .2 Adaptive Fuzzy Control of Robots 机器人的自适应模糊控制,Fuzzy control is the most widely used intelligent control method: The PID fuzzy control, self-organizing fuzzy cont

32、rol, self-tuning fuzzy control, self-learning fuzzy control, expert fuzzy control, etc.,5.4 Intelligent Control of Robot,53,5.4.3 Neurocontrol of Multi-fingered Hands 多指灵巧手的神经控制,Multi-fingered hand is also called multi-joint robot, generally made of a palm and 3 to 5 fingers, while each finger have 3 to 4 joints. Multi-fingered hand is smaller, with more degrees of freedom, and more flexible than normal robot. Multi-fingered hand get much stronger non-linear characteristic than ordinary robot.,5.4 Intelligent Control of Robot,54,5.