1 Introduction
Some time ago a course to learn - industrial robot advanced control technology, the content of the course covers a relatively wide range, although the traditional things in control theory, but there is still a lot of things that we can learn. Especially for those who want to engage in the future direction of the robot, it is particularly important. Recently just I wanted to write a blog column, so tell us about the content of this field.
2. Background
2.1 Development History of industrial robots
- The United States is the birthplace of the modern era of robotics in Japan began with the first robot in 1968. Since then, Japan's industrial robot technology have developed rapidly, and quickly catch up and overtake the United States and in the annual number of installed units, ranking first in the world. The past 20 years, Japan continues to maintain "robot kingdom" status, it has been about 60% of the world's robots
- Although the United States have more than Japanese robot inferior in number (about 15% of the world), but its high technical level, gain some advantages.
- After Germany, the number of robots in the United States station row
- In 1985 our country has already set up a professional committee robot in a few societies, two themes in the late 1980s, the state 863 plan set up in the field of automation, there is about a smart robot.
2.2 basis as a typical industrial robot control object
- Beginning in the 1970s to the late 1980s, is a popular robotics and robot control study period, there were many robotics journals and conferences (such as the Intenational Conference on Robotics and Automation is 87 years from the beginning of each year, held in the United States).
- On the one hand, the most rapid in developing robots that period, bringing the pursuit of high-performance high demand control; on the other hand, control theory at that time has been basically completed the second important development, as modern control theory, optimal control , adaptive control, and coordination of large systems hierarchical decentralized control, robust control, have been basically completed the work to establish the theoretical system, the urgent need to find a typical and influential object system applied research and experimental verification.
- Overall, the study of robot control 70 to late 80s is mainly based on the model, pay attention to the theoretical results; later with the development of intelligent control technology, artificial neural networks, fuzzy control, genetic algorithms have also been applied to robot control but mostly to get simulation and experimental results.
- Complex is a typical robot control object, mainly in complexity:
- a strong nonlinear nature (described low accuracy of the linear model)
- strongly coupled (single loop control method only have a low speed and acceleration control accuracy)
- uncertain parameters, or is difficult to accurately obtain (by challenge model-based control)
and robot objects have a certain representation, such as a kind of mechanical means of open-chain or closed chain. Research needs for high performance robot control method to meet future performance requirements of advanced robots. - There is the "gap" between the control theory point of view, theoretical research and practical application of theoretical developments need to be applied to stimulate demand and drag, the need for complex practical application objects. Research on control theory research robot control does play a role in promoting.
- "Industrial robot apparatus is a machine for the industrial articulated robot of the multi-degree of freedom in the art, the work can be performed automatically, under its own power and the ability to achieve a machine control various functions" --baidu
2.3 Classification robot
Currently the robot can be divided into the following categories, which are listed in the table:
2.4 column focuses on the content
For research institutes and arms control theory of operation is not a new discipline but a comprehensive theory of traditional disciplines:
1, the spatial transform described
•机器人研究关注的空间物体包括：操作臂连杆、末端执行器、操作对象（零部件）等。
•对空间物体用位置和姿态表示其确定状态。
•为描述物体的位姿，需要设置参考系。
•任一坐标系都可以作为描述物体位姿的参考系。
•不同坐标系下的位姿描述可以相互变换。
2、操作臂正运动学
•运动学研究物体的运动，而不考虑引起这种运动的力。
•运动学研究位置、速度、加速度等操作臂运动的几何和时间特性。
•操作臂连杆由关节串接而成。关节有转动和平动两种，关节变量分别为关节角和关节偏距
•自由度：独立的关节变量数。
•基坐标系、工具坐标系。
•正运动学：根据关节变量计算工具坐标系（执行器位姿）。
3、操作臂逆运动学
•给定操作臂末端执行器的位置和姿态，计算对应的关节变量。这是操作臂实际应用中一个基本问题。
•对于开链机构，这是一个复杂的几何问题，会遇到无解（不能到达）或多解问题。
•逆运动学需要求解非线性方程经常没有解析解。
•早期操作臂用人工示教方式记录轨线上的关节变量，省却计算逆运动学关系。
4、速度、静力、奇异性
•前面运动学分析了静态定位问题，这里是速度分析。
•雅可比矩阵给出了从关节速度向笛卡尔空间作业端速度的变换。变换矩阵随操作臂位形不同而改变。
•在奇异位形上，雅可比矩阵不可逆，意味作业端某些速度不可实现。
•雅可比矩阵还可构成关节力矩与作业端对外界施加力/力矩之间的计算关系。
5、动力学
•主要研究产生运动所需要的驱动力。
•动力学方程将关节驱动力矩与路径的空间形式和瞬时特性、连杆和负载的质量特性以及关节摩擦等因素联系起来。
•理想情况下，控制操作臂沿期望路径运动的一种方法是：直接施加运用操作臂动力学方程求解的关节力矩。
•用于仿真。
6、轨迹生成
•将末端执行器的期望运动转化为相应的关节运动，以便实现关节驱动控制。
•一条路径的描述不仅需要确定期望目标，而且还需要确定一些中间点，要求操作臂必须顺序通过这些中间点以便躲避障碍物。
•有时用样条函数来表示一系列路径点的连续函数
7、操作臂设计与传感器
•从控制的角度考察操作臂的机械结构：
–活动范围（包括作业端的位置、朝向）及本体尺寸
–结构的刚性
–机械和控制精度（传动间隙、位置精度、位置重复精度等）
–静摩擦大小
–操作能力（驱动能力、承重比）
•专用操作臂用少的关节完成特定任务，通用操作臂的关节数至少为6。
•完善的操作臂设计还包括：驱动器、传动系统以及内部位置、力传感器等。
8、操作臂运动控制
•基于动力学模型的操作臂控制方法
•控制算法中使用操作臂的位置和速度的量测值，使用方式决定了是线性控制方法，还是非线性控制方法。
•前馈（或补偿）+误差反馈的主流控制结构。
•针对操作臂模型参数未知情况，需要考虑先进的控制技术。
•从“稳定区域”和“误差范围”两个方面比较典型控制方法的性能。
9、操作臂力控制
•运动控制侧重操作端沿给定路径的运动精度控制问题，未考虑操作端与外界环境的接触和作用。
•力控制的应用场合除搬运、各种工具机等，装配也是重要的方面。操作臂的位置控制精度并不高，而装配工艺要求高精度的位控制。因此，工业机器人需要具有力感觉和控制力和力矩的能力，才能弥补位置控制精度不足的缺点而完成装配任务。
•不仅要实现操作端的运动控制，同时还要实现操作端对被操作体的作用力控制，这是一个运动/力混合控制问题。