What is Expert System?

An expert system is a computer program that simulates the judgement and behavior of a human or an organization that has expert knowledge and experience in a particular field. Typically, such a system contains a knowledge base containing accumulated experience and a set of rules for applying the knowledge base to each particular situation that is described to the program. Sophisticated expert systems can be enhanced with additions to the knowledge base or to the set of rules.

Among the best-known expert systems have been those that play chess and that assist in medical diagnosis.

The Journal of Knowledge Engineering publishes papers dealing with all aspects of knowledge engineering, including individual methods and techniques in knowledge acquisition and representation, their application and evaluation, and the construction of systems - including expert systems - based thereon. As well as traditional application areas, such as Software and Requirements Engineering, Human-Computer Interaction, and Artificial Intelligence, we are aiming at the new and growing markets for these technologies, such as Business, Economy, Market Research, and Medical and Health Care. The shift towards this new focus will be marked by a series of special issues covering hot and emergent topics.

Introduction to Robotics

Robot, defined "A re-programmable, multifunctional manipulator designed to move material, parts, tools, or specialized devices through various programmed motions for the performance of a variety of tasks." -- From the Robot Institute of America, 1979

Darrick Addison, an experienced developer in databases, networks, user interfaces, and embedded systems, introduces the field of robotics and the issues surrounding robotic systems. He covers mechanical design, sensory systems, electronic control, and software. He also discusses microcontroller systems, including serial and memory-mapped interfacing, and talks about some of the available open source software options.

The word "robot" originates from the Czech word for forced labor, or serf. It was introduced by playwright Karel Capek, whose fictional robotic inventions were much like Dr. Frankenstein's monster -- creatures created by chemical and biological, rather than mechanical, methods. But the current mechanical robots of popular culture are not much different from these fictional biological creations. Basically a robots consists of:

• A mechanical device, such as a wheeled platform, arm, or other construction, capable of interacting with its environment
• Sensors on or around the device that are able to sense the environment and give useful feedback to the device
• Systems that process sensory input in the context of the device's current situation and instruct the device to perform actions in response to the situation

In the manufacturing field, robot development has focused on engineering robotic arms that perform manufacturing processes. In the space industry, robotics focuses on highly specialized, one-of-kind planetary rovers. Unlike a highly automated manufacturing plant, a planetary rover operating on the dark side of the moon -- without radio communication -- might run into unexpected situations. At a minimum, a planetary rover must have some source of sensory input, some way of interpreting that input, and a way of modifying its actions to respond to a changing world. Furthermore, the need to sense and adapt to a partially unknown environment requires intelligence (in other words, artificial intelligence).

From military technology and space exploration to the health industry and commerce, the advantages of using robots have been realized to the point that they are becoming a part of our collective experience and every day lives.

They function to relieve us from danger and tedium:

• Safety: Robotics have been developed to handle nuclear and radioactive chemicals for many different uses including nuclear weapons, power plants, environmental cleanup, and the processing of certain drugs.
• Unpleasantness: Robots perform many tasks that are tedious and unpleasant, but necessary, such as welding or janitorial work.
• Repetition and precision: Assembly line work has been one of the mainstays of the robotics industry. Robots are used extensively in manufacturing and, more glamorously, in space exploration, where minimum maintenance requirements are emphasized.

Mechanical platforms -- the hardware base

A robot consists of two main parts: the robot body and some form of artificial intelligence (AI) system. Many different body parts can be called a robot. Articulated arms are used in welding and painting; gantry and conveyor systems move parts in factories; and giant robotic machines move earth deep inside mines. One of the most interesting aspects of robots in general is their behavior, which requires a form of intelligence. The simplest behavior of a robot is locomotion. Typically, wheels are used as the underlying mechanism to make a robot move from one point to the next. And some force such as electricity is required to make the wheels turn under command.

Motors

A variety of electric motors provide power to robots, allowing them to move material, parts, tools, or specialized devices with various programmed motions. The efficiency rating of a motor describes how much of the electricity consumed is converted to mechanical energy. Let's take a look at some of the mechanical devices that are currently being used in modern robotics technology.

DC motor: Permanent-magnet, direct-current (PMDC) motors require only two leads, and use an arrangement of fixed- and electro-magnets (stator and rotor) and switches. These form a commutator to create motion through a spinning magnetic field.

AC motor: AC motors cycle the power at the input-leads, to continuously move the field. Given a signal, AC and DC motors perform their action to the best of their ability.

Stepper motor: Stepper motors are like a brushless DC or AC motor. They move the rotor by applying power to different magnets in the motor in sequence (stepped). Steppers are designed for fine control and will not only spin on command, but can spin at any number of steps-per-second (up to their maximum speed).

Servomotors: Servomotors are closed-loop devices. Given a signal, they adjust themselves until they match the signal. Servos are used in radio control airplanes and cars. They are simple DC motors with gearing and a feedback control system.

Driving mechanisms

Gears and chains: Gears and chains are mechanical platforms that provide a strong and accurate way to transmit rotary motion from one place to another, possibly changing it along the way. The speed change between two gears depends upon the number of teeth on each gear. When a powered gear goes through a full rotation, it pulls the chain by the number of teeth on that gear.

Pulleys and belts: Pulleys and belts, two other types of mechanical platforms used in robots, work the same way as gears and chains. Pulleys are wheels with a groove around the edge, and belts are the rubber loops that fit in that groove.

Gearboxes: A gearbox operates on the same principles as the gear and chain, without the chain. Gearboxes require closer tolerances, since instead of using a large loose chain to transfer force and adjust for misalignments, the gears mesh directly with each other. Examples of gearboxes can be found on the transmission in a car, the timing mechanism in a grandfather clock, and the paper-feed of your printer.

Power supplies

Power supplies are generally provided by two types of battery. Primary batteries are used once and then discarded; secondary batteries operate from a (mostly) reversible chemical reaction and can be recharged several times. Primary batteries have higher density and a lower self-discharge rate. Secondary (rechargeable) batteries have less energy than primary batteries, but can be recharged up to a thousand times depending on their chemistry and environment. Typically the first use of a rechargeable battery gives 4 hours of continuous operation in an application or robot.

There are literally hundreds of types and styles of batteries available for use in robots. Batteries are categorized by their chemistry and size, and rated by their voltage and capacity. The voltage of a battery is determined by the chemistry of the cell, and the capacity by both the chemistry and size. See Table 1 for battery sizes.

Table 1. Power supplies

Size	NEDA	IEC	Description
AAA	24A	LR03	Smallest of the command sizes
AA	15A	LR6	Most popular small battery, typically used in packs of 2 or 4
C	14A	LR14	Small flashlight battery, large toys
D	13A	LR20	Largest common battery
9v	1604A	6L-R61	Rectangular with clip-on connector

The robot platform runs off of two separate battery packs, which share only a ground. This way, the motor may dirty up one power source while the electronics can run off of the other. The electronics and the motors can also operate from different voltages.

Electronic control

There are two major hardware platforms in a robot. The mechanical platform of unregulated voltages, power and back-EMF spikes, and the electronic platform of clean power and 5-volt signals. These two platforms need to be bridged in order for digital logic to control mechanical systems. The classic component for this is a bridge relay. A control signal generates a magnetic field in the relay's coil that physically closes a switch. MOSFETs, for example, are highly efficient silicon switches, available in many sizes like the transistor that can operate as a solid state relay to control the mechanical systems.

On the other hand, larger sized robots may require a PMDC motor in which the value of the MOSFET's "on" resistance Rds(on) results in great increases in the heat dissipation of the chip, thereby significantly reducing the chip's heat temperature. Junction temperatures within the MOSFET and the coefficients of conduction of the MOSFET package and heat sink are other important characteristics of PMDC motors.

There are two broad families of transistor: bipolar junction transistors (BJT) and field-effect transistors (FET). In BJT devices, a small current flow at the base moderates a much larger current between the emitter and collector. In FET devices, the presence of an electrical field at the gate moderates the flow between the source and drain.

Sensors

Robots react according to a basic temporal measurement, requiring different kinds of sensors.

In most systems a sense of time is built-in through the circuits and programming. For this to be productive in practice, a robot has to have perceptual hardware and software, which updates quickly. Regardless of sensor hardware or software, sensing and sensors can be thought of as interacting with external events (in other words, the outside world). The sensor measures some attribute of the world. The term transducer is often used interchangeably with sensor. A transducer is the mechanism, or element, of the sensor that transforms the energy associated with what is being measured into another form of energy. A sensor receives energy and transmits a signal to a display or computer. Sensors use transducers to change the input signal (sound, light, pressure, temperature, etc.) into an analog or digital form capable of being used by a robot.

Logical sensors: One powerful abstraction of a sensor is a logical sensor, which is a unit of sensing or module that supplies a particular percept. It consists of the signal processing, from the physical sensor, and the software processing needed to extract the percept.

Proprioceptive sensors: Proprioception is dead reckoning, where the robot measures a signal originating within itself.

Proximity sensors: A proximity sensor measures the relative distance between the sensor and objects in the environment.

Infrared (IR) sensors: Another type of active proximity sensor is an infrared sensor. It emits near-infrared energy and measures whether any significant amount of the IR light is returned.

Bump and feeler sensors: Another popular class of robotic sensing is tactile, or touch-based, done with a bump and feeler sensor. Feelers or whiskers are constructed from sturdy wires. A bump sensor is usually a protruding ring around the robot consisting of two layers.

Microcontroller systems

Microcontrollers (MCUs) are intelligent electronic devices used inside robots. They deliver functions similar to those performed by a microprocessor (central processing unit, or CPU) inside a personal computer. MCUs are slower and can address less memory than CPUs, but are designed for real-world control problems. One of the major differences between CPUs and MCUs is the number of external components needed to operate them. MCUs can often run with zero external parts, and typically need only an external crystal or oscillator.

There are four basic aspects of a microcontroller: speed, size, memory, and other. Speed is designated in clock cycles, and is usually measured in millions of cycles per second (Megahertz, MHz). The use of the cycles varies in different MCUs, affecting the usable speed of the processor. Size specifies the number of bits of information the MCU can process in one step -- the size of its natural cluster of information. MCUs come in 4-, 8-, 16-, and 32-bits, with 8-bit MCUs being the most common size. MCUs count most of their ROM in thousands of bytes (KB) and RAM in single bytes. Many MCUs use the Harvard architecture, in which the program is kept in one section of memory (usually the internal or external SRAM). This in turn allows the processor to access the separate memories more efficiently.

The fourth aspect of microcontrollers, referred to as "other", includes features such as a dedicated input device that often (but not always) has a small LED or LCD display for output. A microcontroller also takes input from the device and controls it by sending signals to different components in the device. Also the program counter keeps track of which command is to be executed by the microcontroller.

R/C Servos: Servomotors, used in radio-controlled models (cars, planes, etc.) are useful in many kinds of smaller robots, because they are compact and quite inexpensive. The servomotors themselves have built-in motor, gearbox, position-feedback mechanisms and controlling electronics. Standard radio control servomotors which are used in model airplanes, cars and boats are useful for making arms, legs and other mechanical appendages which move back and forth rather than rotating in circles.

Animatronic systems Animatronic systems are robotic systems which mimic and look like humans. An android is an anthropomorphic robot -- in other words, a robot that looks like a human.

Pneumatics: Pneumatics is the name for fluid power used in a large number of commercial robots. Pneumatics are also used in a variety of animatronic systems that fall under the category of fluid power. A more well known branch of fluid power is hydraulics. Visit the pneumatics Web site (see Resources later in this article) for additional information.

Open source robot control software

OROCOS (Open RObot COntrol Software) is an effort to start up an open source robot control software project. Broad discussions are being held about what experiences, code and tools can be re-used from other projects, what open standards should be integrated into the project and what organizational structure is most appropriate for the project. Goals of the project are to develop robot control software as follows:

• Under open source and/or free software license(s)
• As modular as possible
• Of the highest quality (from both technical and software engineering perspectives)
• Independent of (but compatible with) commercial robot manufacturers
• For all sorts of robotic devices and computer platforms
• Localized for all programming languages
• Featuring configurable software components for kinematics, dynamics, planning, sensing, control, hardware interfacing, etc.

The project aims to become more than just a copy of existing commercial robot controllers or robot simulation/programming packages. The OROCOS project wants to develop shareable libraries, stand-alone components (sometimes referred to as software agents), and a configurable run-time environment from which to eliminate and control all distributed robotics systems. These types of projects are useful in several ways:

• For re-using code
• For use as an independent sub-system
• For copying their organizational structure
• For learning from the experience of managing an open source project
• For designing and developing extensible and reusable software

Open source matrix libraries

The following are open source matrix libraries that satisfy the above-mentioned requirements. Octave is recommended, since it is GPL-licensed and delivers all required functionality.

GNU Octave: GNU Octave is a high-level language, primarily intended for numerical computations. It provides a convenient command line interface for solving linear and nonlinear problems numerically, and for performing other numerical experiments using a language that is mostly compatible with Matlab. It is easily extensible and customizable via user-defined functions written in Octave's own language, or using dynamically loaded modules written in C++, C, Fortran, or other languages.

GNU Octave is freely distributed software. You may redistribute it and/or modify it under the terms of the GNU General Public License (GPL) as published by the Free Software Foundation. More detailed information about GNU Octave can be found by visiting the Octave Web site (see Resources).

GSL (GNU Scientific Library) GSL is an ongoing effort to develop a modern extensive and ANSI C library for numerical computing. The GNU Scientific Library (GSL) is a collection of routines for numerical computing. The routines are written from scratch by the GSL team in ANSI C, and are meant to present a modern Applications Programming Interface (API) for C programmers, while allowing wrappers to be written for very high level languages.

GSL is free software. It is distributed under the terms of the GNU General Public License. Visit Red Hat's Web site (see Resources) for more information concerning GSL.

Real time kernels

Real-Time Linux (RTLinux) RTLinux(TM) is a hard real-time operating system that handles time-critical tasks and runs Linux as its lowest priority execution thread. In RTLinux, the kernel shares one or more processors with standard Linux. This allows the system to run accurately timed applications performing data acquisition, systems control and robotics, while still serving as a standard Linux workstation. Version 3.0 (final) is available on the Web at ftp.rtlinux.com (see Resources).

RTLinux.org is the non-commercial RTLinux site for the open source user and developer community. Their sister site, RTLinux.com (see Resources), discusses commercial support and development.

eCos (embedded Configurable operating system): eCos is an open source real-time operating system for deeply embedded applications. It meets the requirements of the embedded space that Linux cannot yet reach. Linux currently scales upwards from a minimal size of around 500 kilobytes of kernel and 1.5MB of RAM, before taking into consideration application and service requirements. The eCos open source project can be found at their Web site (see Resources).

RTEMS (GPL License): RTEMS is an open source real-time operating system and environment for C, C++ and Ada95. It is distributed under the terms of the GNU General Public License.

Visit the RTEMS site (see Resources) for downloads and more detailed information about RTEMS.

Utilities and tools

ROBOOP (A robotics object oriented package in C++): This package is an object-oriented toolbox in C++ for robotics simulation. Technical references and downloads are provided in the Resources.

CORBA: A real-time communications and object request broker software package for embedding distributed software agents. Each independent piece of software registers itself and its capabilities to the ORB, by means of an IDL (Interface Definition Language). Visit their Web site (see Resources) for technical information, downloads, and documentation for CORBA.

TANGO/TACO: This software might be useful for controlling a robotics system with multiple devices and tools. TANGO is an object oriented control system based on CORBA. Device servers can be written in C++ or Java. TACO is object oriented because it treats all (physical and logical) control points in a control system as objects in a distributed environment. All actions are implemented in classes. New classes can be constructed out of existing classes in a hierarchical manner, thereby ensuring a high level of software reuse. Classes can be written in C++, in C (using a methodology called Objects in C), in Python or in LabView (using the G programming language).

TACO was designed to be portable and runs on a large number of platforms (for example, Linux, Solaris, HP-UX, Windows/NT, Windows/95, and OS9). To download the source code and other technical documents visit their web site (see Resources).

Controllers

Task Control Architecture: The Task Control Architecture (TCA) simplifies building task-level control systems for mobile robots. "Task-level" refers to the integration and coordination of perception, planning, and real time control to achieve a given set of goals (tasks). TCA provides a general control framework, and is intended to control a wide variety of robots. TCA provides a high-level machine-independent method for passing messages between distributed machines (including between Lisp and C processes). TCA provides control functions, such as task decomposition, monitoring, and resource management, that are common to many mobile robot applications. The Resources section provides technical references and download information for Task Control Architecture.

EMC (Enhanced Machine Controller): The EMC software is based on the NIST Real time Control System (RCS) methodology, and is programmed using the NIST RCS Library. The RCS Library eases the porting of controller code to a variety of UNIX and Microsoft platforms, providing a neutral application programming interface (API) to operating system resources such as shared memory, semaphores and timers. The EMC software is written in C and C++, and has been ported to the PC Linux, Windows NT, and Sun Solaris operating systems.

Darwin2K: Darwin2K is a free, open source toolkit for robot simulation and automated design. It features numerous simulation capabilities and an evolutionary algorithm capable of automatically synthesizing and optimizing robot designs to meet task-specific performance objectives.

Languages

RoboML (Robotic Markup Language): RoboML is used for standardized representation of robotics-related data. It is designed to support communication language between human-robot interface agents, as well as between robot-hosted processes and between interface processes, and to provide a format for archived data used by human-robot interface agents.

ROSSUM: A programming and simulation environment for mobile robots. The Rossum Project is an attempt to help collect, develop, and distribute software for robotics applications. The Rossum Project hopes to extend the same kind of collaboration to the development of robotic software.

XRCL (Extensible Robot Control Language): XRCL (pronounced zircle) is a relatively simple, modern language and environment designed to allow robotics researchers to share ideas by sharing code. It is an open source project, protected by the GNU Copyleft.

Open System Architecture for Controls within Automation Systems (OSACA): OSACA is a joint European project that aims to improve the competitiveness of the manufacturers of machine tools and control systems in the world market. The main goal of the project is to specify system architecture for open control systems, which is manufacturer independent.