Current Situation and Prospect of robot controller

1. Introduction

It has been 50 years since the birth of the first remote control robot in the world. In just a few years

With the development of theories and the needs of industrial production and the advancement of related technologies, robotics have gone through three generations [1]: (1) programmable display

Teach reproduction robots; (2) control robots with certain autonomous capabilities based on sensors; (3) intelligent robots as robots

The robot controller is one of the key parts that affect the performance of the robot. It affects the development of the robot to a certain extent.

Previously, due to the great progress in artificial intelligence, computer science, sensor technology and other related disciplines, robot research has become a high level

And higher requirements for the performance of the robot controller.

For different types of robots, such as walking robots with legs and industrial robots with joints, the overall method of the control system is relatively poor.

The controller design scheme is different. This article only discusses the issue of industrial robot controller.

2. robot controller type

A robot controller is a device that controls a robot to perform certain actions or tasks based on commands and sensing information. It is a robot.

Determines the performance of the robot.
Robot Control Algorithms can be divided into two types: Serial and parallel structures.

2.1 serial Processing Structure

The so-called serial processing structure means that the robot's control algorithm is processed by a serial machine.

Structures and control methods can be divided into the following types [2].
(1) single CPU Structure and centralized control mode
A powerful computer is used to implement all the control functions. In earlier robots, such as hero-I and robot-I

This structure requires a lot of calculations (such as coordinate transformation) in the control process. Therefore, this structure is slow.
(2) second-level CPU Structure and master-slave control mode
The first-level CPU is the host and serves as the system management, robot language compilation, and human-machine interface functions. It also uses its computing capability to complete

Coordinate Transformation and trajectory interpolation, and regularly send the calculation results as the increment of joint motion to the public memory for the second-level CPU to read; second-level

The CPU completes digital control of all joint locations. there is basically no connection between the two CPU buses of this type of system, only through public memory swap

It is a loosely coupled relationship. It is very difficult to use more CPU to further distribute the function. Japan produced the motoman in 1970s.

The computer system of the robot (5-joint, DC Motor Drive) belongs to this master-slave structure.

(3) multi-CPU Structure and Distributed Control Mode

Currently, this level-2 distributed structure is widely used, and the upper and lower computers are responsible for the whole system management, as well as the kinematics calculation and trajectory.

Planning, etc. The lower computer is composed of multiple CPUs, each CPU controls a joint movement, These CPUs and the master machine contact is in the form of a bus tight

Coupling. This structure significantly improves the speed and performance of the controller. However, the Features shared by these multi-CPU systems are all for specific purposes.

The distributed structure of functions, that is, each processor undertakes a fixed task. Currently, most of the commercialized Robot controllers in the world are

This structure.
The location control part of the controller computer control system uses digital location control almost without exception.

These controllers use serial machines to compute robot control algorithms. They share a common weakness: Negative computing.

Because of its heavy workload and poor real-time performance, most of them adopt offline planning and feed-forward compensation decoupling to reduce the computing burden in real-time control.

When a person is disturbed during operation, the performance will be affected, making it difficult to guarantee the accuracy indicators required in high-speed motion.

Due to the complexity of robot control algorithms and the urgent need to improve the robot control performance, many scholars have made improvements in modeling and algorithms.

But it is still difficult to meet the real-time computing requirements on the serial structure controller.

One method is to use a high-grade microcomputer or minicomputer. The other is to use a multi-processor for parallel computing to improve

The computing power of the controller.

2.2 Parallel Processing Structure

Parallel processing technology is an important and effective way to improve the computing speed and meet the real-time requirements of Robot Control.

Looking at the parallel processing technology of Robot controllers, we have studied the parallel algorithms and their implementations of robot kinematics and dynamics.

In 1982, J. Y. S. luh [3] proposed for the first time the parallel processing of robot dynamics, because the dynamic equation of the articulated robot is

A group of nonlinear strongly coupled second-order differential equations, the computation is very complex, and the speed of robot dynamics algorithm is also to achieve complex control.

Algorithms such as Torque Calculation, nonlinear feed-forward method, and adaptive control method lay the foundation for developing parallel algorithms.

The serial algorithm is used to make it parallel and then map the algorithm to the parallel structure. Generally, there are two methods: first, consider

Based on the computing model supported by the processor structure, develop the parallelism of the algorithm. Second, first develop the parallelism of the algorithm, and then design

The parallel processor structure of the algorithm is used to achieve optimal parallel efficiency.

The computer system that constructs a robot controller with a parallel processing structure generally adopts the following methods:

(1) Development of specialized ASIC for Robot Control [4, 5]

The design-specific large-scale architecture makes full use of the parallelism of robot control algorithms, and is easy to solve Robot Control Problems Based on the parallel architecture in the chip.

A large number of calculations in the algorithm can greatly improve the calculation speed of the kinematics and dynamic equations. However, because the chip is based on the specific algorithm

When the algorithm is changed, the chip cannot be used. Therefore, the controller constructed in this way is not universal and is not conducive to the System

Maintenance and Development.
(2) using a chip computer with parallel processing capabilities (such as Transputer and DSP) to form a parallel processing network

Transputer is a kind of chip computer for Parallel Processing developed and produced by British inmos.

It is easy to construct different topology structures, and Transputer has strong computing power.

Transputer parallel processor, which is used to construct a variety of robot parallel processors, such as pipeline type and tree type.

The Transputer network implements inverse kinematics computation. For the purpose of real-time control, the article [7] implements feed-forward compensation and Torque Calculation respectively.

Two control schemes based on fixed models.

With the increasing speed of digital signal chips, high-speed digital signal processor (DSP) has been widely used in information processing.

Application. DSP is a fast digital computing speed and easy to form a parallel processing network [8]. Document [9] describes a DSP-based

The robot controller adopts the parallel/pipeline design scheme to improve the controller performance.

(3) Use a general microprocessor

The general microprocessor is used to construct a parallel processing structure that supports computing and online real-time computing of complex control policies.

[10, 11.

3. Problems with robot controllers

With the rapid development of modern science and technology and the progress of society, the performance of robots is more demanding.

It has become the main development direction in the robot field, such as Precision assembly robots, force/position hybrid control robots, and multi-body coordination control

As well as the research of robots in Advanced Manufacturing Systems. Correspondingly, the performance of Robot controllers is also put forward higher requirements.

However, since its birth, the controllers used by robots, especially industrial robots, are basically independent by developers.

This architecture is developed using specialized computers, specialized robot languages, specialized operating systems, and specialized microprocessor.

The controller cannot meet the requirements of modern industry development.

From the two types of Robot controllers mentioned above, the serial processing structure controller is closed, with a single function and computing power.

Poor, difficult to ensure real-time control requirements, so the vast majority of commercial robots are currently using single-axis PID control, difficult to meet the machine

High-speed and high-precision requirements of human control. Although the distributed structure is open at a certain level, you can add more

Processor to meet the needs of sensor processing and communication, but it is only available within a limited range. As mentioned in [12]

For example, Puma robot uses PDP-11 as the upper-level master computer), specialized robot language (such as Val ),

The microprocessor and the control algorithm are fixed in the EPROM "structure limits its scalability and flexibility, so its structure is closed.

Although the parallel processing structure controller can make great breakthroughs in computing speed and ensure real-time control, we must look at it.

There are still many problems. The current research on parallel processing controllers is generally focused on parallel processing of robot kinematics and dynamic models.

It is designed based on the ing features of parallel algorithms and multi-processor structures. That is, by decomposing a given task, several subtasks are obtained and listed.

The data-related flow chart enables concurrent processing of each sub-task on the corresponding processor. Due to the inherent features such as communication and synchronization in parallel algorithms,

If the program is improperly designed, locking and communication congestion may easily occur.

In combination, the existing robot controller has many problems, such:

(1) Poor openness

Limited to the closed structure of "dedicated computer, dedicated robot language, dedicated microprocessor ".

It has specific functions and is suitable for specific environments and is not easy to expand or improve the system.

(2) Poor software independence

The software structure and its logical structure depend on the processor hardware, and it is difficult to transplant between different systems.

(3) Fault Tolerance

Due to the inherent characteristics of data relevance, communication, and synchronization in parallel computing, the Controller's fault tolerance performance becomes worse.

Failure may cause paralysis of the entire system.

(4) poor scalability

Currently, Robot controllers focus on improving and improving system performance at the joint level. Due to the closed structure, it is difficult

To expand the system as needed, such as adding function modules such as sensor control.

(5) The network function is missing

Currently, almost all robot controllers do not have network functions.

In general, the robot controllers in both serial and parallel structures are not open architectures

Or hardware is difficult to expand and change. For example, the Controller of the commercialized motoman robot is not open, and it is difficult for users

It is necessary to modify and expand its functions. The common practice is to perform a detailed Anatomical analysis on it and then transform it.

4. Prospect of robot controller

With the development of robot control technology, in view of the defects of Robot controllers with closed structures, the development of "modular with open structures"

Standardized robot controller is a development direction of robot controller. In recent years, Japan, the United States and some European countries

All are developing open-architecture Robot controllers, such as the open-architecture and network functions developed by yaskawa Corporation in Japan based on PC

Robot controller [13]. The subject of China's 863 plan intelligent robot has also established research projects in this area [14].

Open Structure robot controller refers to: the controller design layers are open to users, users can easily expand and improve

The main idea of its performance [15] is:
(1) Use development systems based on non-closed computer platforms, such as Sun, SGI, and PC's.

Hardware and software resources create conditions for controller expansion.
(2) Use standard operating systems, such as UNIX, vxwork, and standard control languages, such as C, C ++. Use Standard Operating Systems

And control language, which can change the coexistence and incompatibility of various specialized robot languages.
(3) using a standard bus structure, hardware required to extend the controller performance, such as various sensors, I/O Boards, and motion

The control panel can be easily integrated into the original system.
(4) use network communication to achieve resource sharing or remote communication. Currently, almost all controllers do not have network functions.

Network Communication can improve the flexibility of system changes.
We can design Robot controllers with open structures according to the above ideas, and implement modules as much as possible during the design process.

Modularization is a modern method of system design and establishment. The system is designed according to the modular method and consists of multiple functional modules.

The system established in this way is not only of good performance, short development cycle, but also low cost. modularization also makes the system open.

Easy to modify, reconstruct, and add configuration functions. [16] The new controller is based on the concept of multi-Self-subject.

The system consists of the database module, communication module, sensor information module, human-machine interface module, scheduling module, planning module,

The servo control module consists of seven modules.

The new robot controller should have the following features:

(1) Open System Structure

The open software and hardware architecture can be used to expand functions as needed, so that they can be applied to different types of robots or robots.

Automatic production line.

(2) Reasonable Modular Design

For hardware, modular design based on system requirements and electrical characteristics not only facilitates installation and maintenance, but also improves the system

And the system structure is more compact.

(3) effective task division

Different subtasks are implemented by different function modules to facilitate modification, addition, and configuration of functions.

(4) Real-time and multi-task requirements

The robot controller must be able to process external interruptions within a specified period of time, and can simultaneously perform multiple tasks.

(5) Network Communication

The network communication function is used to facilitate resource sharing or collaboration among multiple robots.

(6) Visual Human-Machine Interface

In addition, the motion control panel is essential in the robot controller. Due to the different performance of the robot

The PMAC (Programmable Multi-axies Controller) launched by Delta Tau in the United States

PMAC is a powerful motion controller. It fully develops the powerful functions of DSP technology and provides users

With the help of Motorola's dsp56001 digital signal processor, PMAC can simultaneously operate

1 ~ The 8-axis has many advantages over other motion control panels.

Due to the wide variety of software and hardware suitable for Robot Control and the rapid development of modern technology, the development of a fully open structure

The quasi-robot controller has some difficulties, but the existing technologies, such as the openness, security, networking, and standard of Industrial PCs, are used.

The real-time multi-task operating system, standard bus structure, standard interface, etc., breaks the existing robot controller structure closed situation, open

Standardized Robot controllers with open structures and modular functions are completely feasible.

5 conclusion

With the development of robot technology and the continuous expansion of the robot application field, it puts forward higher requirements on robot performance. Therefore,

How to effectively apply the research results of other fields (such as processing, voice recognition, optimal control, and AI) to robots

The real-time operation of control systems is a challenging research work, and modular and standardized robots with open Structures

The Research on controllers is undoubtedly of great significance for improving robot performance and autonomous capabilities and promoting the development of robot technology.