Robotic Systems
A robot, (from the Czech word robota 'slave work, drudgery'), is
a programmable machine that can perform autonomously a number of
tasks by interacting with its surroundings. Robots can, in some
cases, replace human labour; the robot can, for instance, move
materials and manipulate objects faster, cheaper and more precisely
than human beings, and that is why robots are widely used by
industry.
The word robot was introduced by the Czech artist Josef Capek
and popularised through his brother Karel Capek's play R.U.R.
(1920).
Behaviour
A robot's behaviour differs from that of a computer programme in
the sense that it interacts with its physical surroundings by means
of sensors and effectuators (often motors). Sensors allow the robot
to sense external effects from its surroundings, while effectuators
allow the robot to influence and manipulate its surroundings. In
addition, autonomous (self-controlling) robots require their own
energy source, e.g. solar cells or batteries, and their own control
system in the form of a control programme placed in the robot
itself.
Sensors
Sensors for robots include light sensors e.g. photo resistances,
IR distance detectors, pyrometers and cameras; power sensors, e.g.
switches (pressure sensors), flex sensors and power measuring
resistances; sound sensors, e.g. sonar, microphones and speak
recognition circuits; position and orientation sensors, e.g.
revolution counters, compasses, gyro compasses and gradient sensors
plus internal condition sensors that are able to measure the level
of the battery or the temperature of the robot.
Effectuators
Effectuators allow a robot to influence its surroundings. The
impact may happen through movements of the robot itself, which is
the case when mobile robots move around by means of wheels, legs or
caterpillar tracks. Generally speaking, influence is provoked by
manipulators that may be stationary or mobile, and which are used
e.g. for moving materials and tools. The development of machines
able to manipulate its surroundings played a major role in the
automation processes that streamlined factories in the seventeen
and eighteen hundreds.
Control
There are several methods for controlling robots. One of these
methods stems from the research areas "artificial intelligence" and
"cybernetic", in which scientists have studied robots as an
opportunity to create intelligent machines. One of the big debate
topics in the field of artificial intelligence is whether
intelligence is viable without a physical body.
The scientists that perform research into embodied artificial
intelligence maintain that intelligence must be studied as a
physical system placed in real surroundings. This direction within
artificial intelligence has been one of more pacesetters in the
development of methods for designing robotic control systems. Such
methods include classical robotic technologies as well as novel
methods that are inspired by natural systems.
Classical control
Within classical robotics, scientists distinguish between
various forms of control: "open loop", "closed loop" and
"feedback". Open-loop control allows the robot to move according to
a predetermined pattern without regard for the actual behaviour of
the robot, while closed-loop control gives the robot an opportunity
to recognise its actual behaviour. This ability is used in feedback
control for instance, where the behaviour of the robot is
influencing the control system that controls it so that a change in
the robot's behaviour may lead to a new kind of control.
The classical negative feedback control, "PID control"
(proportional-integral-derivative), is an example of this. Here the
control is changed subject to feedback in terms of position errors
(proportional), state-of-balance position errors (integral) and
velocity errors (derivative). Robots with a classical control
system can be very precise and that is why they are widely used by
industry even though they are often heavy, slow and minimally
autonomous.
Behaviour-based robots
Since mid 1980s scientists have developed behaviour-based robots
as a potential solution to the problems of classical robots such as
weight, velocity and autonomy. Often, however, the behaviour-based
robots are less precise than the classical robots. They were
introduced in 1986 by Rodney Brooks (b. 1954), an American from
MIT, USA, via the so-called "subsumption"-architecture, as a
reaction against classical robotics; subsequently this form of
robotics has moved towards its own science area.
The behaviour-based robots are characterized by decentralized
control where the robot control system is divided into modules that
each provides a behaviour that runs in parallel. In addition, some
behaviour modules are designed to provide a reactive behaviour to
the effect that a sensing input makes the modules react with a
reflex motor movement. Since the modules run in parallel the result
is faster control, and often the robots are robust such that do not
stop even if a minor part of the control system.
Behaviour-based robots have given inspiration to the integration
of new robotic control-system methods based on autonomy and the
desire to equip robots with an assimilation capacity of their
own.
Such control systems comprise artificial neural networks and
development by means of "reinforcement learning" and evolutionary
methods. By using these methods it is possible to obtain adaptive
robots that have a certain ability to acquire new knowledge while
they operate in their own surroundings. It is done by teaching the
control system to learn new reaction patterns based on the
situations that the robot may be exposed to. By using this type of
methods scientists try to solve some of the problems connected with
classical robots viz. the robots can often only move in
pre-programmed tracks and they are extremely sensitive to changes
in the surroundings.
It is possible for adaptive robots to operate in different
surroundings and to learn about changes in such surroundings so
that they can cope with certain unforeseen and non-pre-programmed
events.
Applications
Artificial neural networks used as control systems for robots
are inspired by natural systems and natural nerve systems. Hence,
one research area within artificial life applies robots for
biological examinations e.g. in behaviour biology. In such tests a
robot is used to implement hypotheses about an animal and
subsequently to test a hypothesis by placing the robot under the
same experimental conditions as the test animal.
Intensive research is performed into how such adaptive methods
can be used in robots that are to explore unknown or difficult
accessible areas (e.g. mining areas, the core of nuclear reactors,
submarine areas, planets).
Classical robots are widely used in repetitive, non-dynamic work
in industry e.g. assembly-line work, welding and spray painting. In
addition, robots are used for a number of surgeries; research is
also performed into the development of robots that can assist the
elderly and the disabled e.g. as robot guides for the blind and as
autonomous robot wheelchairs.
Entertainment industry
Robots are widely used in the entertainment sector, where toy
robots are used as construction toys and electronic pets. Besides,
robots are often used in movie pictures.
Science fiction
The popular perception of robots is especially influenced by the
science fiction literature. The American writer and biochemist
Isaac Asimov was prominent in the introduction of robots in the
1950s and as the originator of the three ethical laws on robotics
(also known as Asimov's laws). In the real world robots differ
considerably from robots i science fiction which often demonstrate
human capabilities and character traits. In contrast, the real
robots' sensing, action and energy-producing capabilities are very
limited in comparison with the corresponding human
capabilities.
Source: Professor, Henrik Hautop Lund, Tecnical
University of Denmark