| Robotics |
Above: The Cybex Warbot 7000 series developed by the Mechatronics Corp in collaboration with the
Cyberprime and Borgtech Corporations, click to enlarge. This is one of the most advanced weapon
systems of its kind in the known Universe. It has extreme mobility, and is capable of running at sustained
speeds of 100 kilometres per hour. Powered by an EPA antimatter 8X series generator the Cybex can
metres tall, the Cybex 7000 is extremely agile and its excellent fire control systems ensure near 100%
accuracy at ranges up to one kilometre. The actuators are operated by advanced polymeric
syntho-muscle fibres which give the Cybex the strength (though not the inertia) of some 20 men.
The Cybex 7000 has a quantum positronium CPU (QX 7990) which is only some 9 cm in diameter but is
capable of advanced and rapid parallel computations for battle-field situations. Some components of this
system require freezing to -100 degrees Celsius for optimum performance and the antimatter generator
can also generate large amounts of heat. In addition rapid movement and use of energy weapons all
generate surplus heat which must be dissipated, however, it is important to maintain a low heat signature
for stealth. To achieve this the Cybex 7000 has a thermal screen which converts heat and infrared
radiation into higher frequency electromagnetic energy, either optical light or ultraviolet. This light is
radiated through a crypsis screen which can match the outgoing light to the environment, and thereby
providing cryptic camouflage at the same time as radiating away excess heat energy whilst maintaining a
faint infrared signature. The crypsis screen can also absorb, scatter or reflect hostile laser beams, and
other heat attacks, minimising damage from laser hits.
Other defensive systems include a negative screen which provides protection against electrical attacks
and repels anyone who attempts to grapple the Cybex 7000 by administering a severe electrical shock on
contact. The microarchitecture of the armour system is designed to channel frequencies of ultrasound
commonly used in sonic beam weapons around the Cybex, so that the ultrasound passes without causing
damage. The armour itself is made up of tough composite materials to absorb kinetic energy impacts,
including a layer which dissipates shock waves that travel through the armour, and thereby protecting the
delicate internal components.
The Cybex 7000 can be equipped with a variety of weapons and tools, the one shown here is armed with 4
Androm X-9 particle cannons (PCs) on its head turret, which can alternate between pulsed or beam mode
and between negatively and positively charged particles. These cannons have a maximum effective range
of 500 metres. The right arm is carrying an Omega-5 Mk 6 variable wavelength armour-piercing laser
cannon (APLC), with an effective maximum range of 3 kilometres. These weapons can also be operated in
laser lance mode for drilling through especially tough armour plate, such as on the hull of a warship. The
left arm is carrying the Battletex Epsilon-theta Mk 5 heavy ion cannon (HIC) which can operate in beam
mode or in wide-field mode when used to disrupt communications and electrical equipment.
The principles of robotics
We shall compare robotic systems to living systems with particular reference to what is possibly the most
sophisticated robot so far developed by Earthlings, ASIMO developed by Honda, and the human being.
Search for ASIMO in Google or on Howstuffworks.com.
1. Power
Providing energy for robots is no small task. The average human male uses about 120 watts of power
averaged over the course of a day, with peak outputs reaching around 1000 W during sprinting. As a
comparison, 120 W is the typical power provided by a one square metre solar panel, and the power of the
Sun's rays striking the Earth's atmosphere is 1366 W per square metre. The most advanced robots built
by Earthlings to date, include the Honda ASIMO humanoid robot is powered by a 51.8 V lithium-ion battery
weighing 6 kilos, which provides enough energy for one hour before it needs recharging and ASIMO has
only the fraction of the athletic power of a real human.
2. Strength
ASIMO can only lift a few kilos in weight. However, human-sized androids have been developed in Japan
that can allegedly lift around 70 Kg. You might think that it was an easy task to give a robot strength, after
all heavy hydraulic machinery is enormously strong, however these systems do not scale down well to the
size of a human being. ASIMO uses 34 servo motors, or electric motors that work by rotating shafts. Such
motors are clearly power hungry. Human muscles contains arrays of protein filaments (polymers, molecular
motors) that slide together using chemical energy, causing the muscle to shorten (animal muscles always
pull and never push). Similarly, robot muscles may operate using polymer fibres that change their
conformation or shape, perhaps by coiling into tight helices when an electric current is applied, similarly
causing the muscle fibres to shorten. However, making a 100 kg robot that could lift 250 kg (and thus
match some of the strongest humans) is no easy task.
3. Agility
ASIMO movements are limited. Each of its joints is operated by an electric servo motor, of which there are
34 in all. Humans, in contrast, the human body has between 565 and 850 separate muscles (the exact
number varies between individual and some muscles may merge and so appear as a single muscle). This
gives the human body a much greater range of motion. However, powering 800 electric motors would
require considerably better batteries! Again, molecular motors may be a better solution. However, ASIMO
can walk down steps, run in a circle and lift objects without breaking them.
4. Speed
ASIMO can run at 6 kilometres per hour (compared to a jogging speed of about 13 kilometres per hour for
a reasonably fit human). In humans, the thickness of the muscle as well as the muscle type determines the
muscle's strength. However, longer muscles and longer limbs are faster, all else being equal. This is why
sprinters tend to be tall and javelin throwers tend to be tall to give their long arms speed. However, shorter
limbs are stronger, since they have better leverage, so we would have short and thick limbs for strength
and long limbs for speed. Power is another parameter, and is equal to the rate at which the muscles can
do work (the maximum rate at which they can expend useful energy) and this is proportional to the total
volume of the muscle, so for maximum power we would have long and thick limbs. This is one reason why
athletes from different sports are so different in body shape, especially at the elite end of the spectrum.
Muscles also come in two main types - one that is fast and strong but fatigues quickly, and so is useful for
sprinting, and one which is slower and weaker but fatigues much more slowly, and so is good for distance
running. The proportions of each muscle type that a human has depends largely upon genetics - some
people innately make better sprinters or weight-lifters whilst others make better marathon runners. In the
end it is impossible to maximise speed, strength, power and speed - at some point there is a trade-off and
one has to compromise or invest most heavily in whichever is most important for the specific purpose.
5. Weight
Humans are often conditioned into thinking that the heavier the better - since weight is good for pushing
smaller creatures out of the way. However, a larger body requires more energy to move and maintain and
more raw materials. If you look at a range of animal species working in Earth's gravity, then you will find
that the fastest animals tend to be about 100 kg in weight and smaller and larger animals tend to be slower
(though there are exceptions of course). Smaller animals lack the muscle power whilst larger animals
simply have too much bulk to move (though an elephant can still outrun a human over short
distances)Humans are actually rather slow animals with the fastest humans ranking mediocre in sprinting
speed. However, humans make very good distance runners (with training!). Having less weight to carry
certainly means that you use less energy when travelling over a greater distance. Constructing robots from
materials which are light and strong is a major goal. Lighter materials will mean that less power is needed
to move the robot. Sometimes weight is an advantage - when it comes to crashing through concrete walls,
for example, and some warbots are heavily built for this reason, just like heavy tanks. However, even in
main battle tanks, weight reduction is an issue - lighter armour means that more weight can be carried in
munitions, for example. Animals (and plants) are made out of materials that are enormously strong but
very light. Bone is stronger than steel reinforced concrete, but is a fraction the weight. The cellulose and
collagen that make up plant tissues are strong as or stronger than steel, but a fraction of the weight.
Carbon-based materials (including animal and plant tissue components) make very strong structures, such
as silk protein, but are much lighter than steel. Proteins have a consistency rather like plastic, but can be
much stronger. Spider silk has a strength comparable to synthetic carbon fibres or kevlar, but can be
made by the spider at room temperature and pressure, using natural and renewable resources.
6. Mental processes
So far the hardest task for robotics engineers on earth has been producing robot brains. ASIMO carries
two laptop computers in its backpack, but still requires a human controller with a joystick, although it
performs the details of its instructed tasks automatically, but it cannot think! Incorporating powerful
computers will also increase the power demands of a robot. The human brain has a power output of about
20 watts - it weighs less than 2% of the body's total weight, but consumes some 20% of the body's energy.
Current estimates place the processing power of the human brain at between 10^15 and 10^20 operations
per second, though there are reasons to expect that this could be 1000 times as great as this. This
compares to the Blue Gene supercomputer, the fastest on Earth with 3.6^14 operations per second
(strictly floating point operations per second, or flops per second). Such computers can simulate a
simplified mouse brain at one tenth normal speed, suggesting that the human brain's processing speed is
certainly nearer the upper end of the estimates given (say 10^16 operations per second or higher). The
Blue Gene has 2^16 compute nodes and some 64 cabinets. This compares with the human brain with
10^11 neurones (brain cells), each with some 1500 synapses on average (though some neurones may
have 10 000 synapses or so) gives the brain 10^14 to 10^15 compute nodes, but each is capable of
performing 10 operations per second (and possibly up to 1000), enabling the brain to perform say 10^16
operations per second. However, in addition to these neurones, there are ten to fifty times as many
neuroglial cells in the brain, some of which appear to have (unknown) functions in processing, enabling the
brain to perform perhaps 10^19 operations per second. There are other possibilities that may further
increase this number 10-100 hundred times. Thus, the Blue Gene supercomputer is probably some
100-1000 000 times short and it would take 100 Blue Genes to stand any chance at all of matching the
human brain and that's at least 6 400 cabinets consuming massive amounts of electricity (26 800 watts per
cabinet or 1.7 million watts in total!) compared to a 1.5 lb brain consuming 20 W! Nevertheless, it does
look within the grasp of Earthling technology to match this complexity soon. A typical desktop computer is
way behind at around 10^9 operations per second.
Quantum computers (such as that used by the Cybex 7000) have the potential to thrash the Blue Gene
supercomputer and use much less power and occupy much less space, but Earthlings will have to wait and
see how effective these quantum computers can be. At this point you might be thinking why is it then that
the Blue Gene can perform computations vastly more complex than what you can do in your head. The
answer is simply that most of the brain's function is automatic and subconscious and so you do not have
much control over it. That isn't to say that the brain is reckless, rather most of the brain is involved in
calculations concerning internal organs and body maintenance. Also when you catch a thrown ball, you do
not consciously compute velocities, trajectories and angles in order to compute your own trajectory to
intercept the ball - this is done for you, automatically by the vast computational power of your brain.
Programming ASIMO to do the same tasks is proving immensely difficult! Even the tiny brain of an insect
has some one million neurones and so may perform an estimated 10^11 to 10^16 operations per second,
still equivalent to a thousand or so desktop PCs!
7. Intelligent materials
Living organisms are made from intelligent materials, by which i mean materials that respond to
environmental stress, modifying themselves when they need to improve and repairing themselves when
damaged. Robots too can be constructed from intelligent materials. Sensors can detect the stresses and
strains placed on joints, limbs and motors and skin sensors can detect damage. Pain is a necessity for
conscious organisms. If you did not feel pain then you would not take all the necessary steps to avoid
damage. Some people are born without the genetic ability to perceive pain and these people die young. In
this way, pain is a useful survival aid (though like all systems it can go wrong in the case of useless and
debilitating chronic pain). Pain is often suppressed when an organism is in imminent danger - there are
many reports of soldiers, for instance, not feeling their wounds until help had arrived. Even a tough warbot
must be programmed to avoid damage (indeed warbots especially so) but we do not want inappropriate or
debilitating responses. If damaged, materials can be designed to repair themselves. For example, liquid
chemical constituents can be pumped to the damaged area where they can solidify to form a protective
scar, or if more advanced, they may self-assemble into new materials of the right type. Tiny robots
(possibly nanobots) inside the main robot can also assist with repairs, though there are other conceivable
methods. Think how short lived most of the machines around you are (cars and computers etc.) that is
partly because they are unable to repair even daily wear and tear. It is also because biological materials
are very durable - even if your bones could not repair (and they certainly do!) they would last some 20
years of normal usage. (Furthermore, bones actually get stronger with increased usage). The brain can
also respond with damage and wear by re-routing its circuits to make maximum use of available brain
tissue. If your desktop blows its CPU, then it is defunct. The Cybex 7000 incorporates repair and
maintenance systems and can remain active for decades, perhaps even centuries. The Cybex 7000 also
has the ability to reroute damaged circuitry in its central processor, so that quite extensive damage is
required to completely put it out of operation.
8. Sensors
If you took the eyes of an eagle, the nose of an elephant, the ears of a bat (for sake of arguments) and
the ability of a goldfish to see in infrared and ultraviolet as well as in colour, and the sense of direction of a
homing-pigeon and you would have a very sensitive animal indeed! However, sensors are only as good as
the computer that processes the information they receive. Processing information received from the eyes
of a human being take up a large part of the brain, whilst the sense of smell takes up only a small part in
humans (but a large part in dogs) so to have all senses highly tuned would result in an excessively large
brain! Nevertheless, animals are extremely sensitive - their eyes can detect a single photon of light.
Background noise is another limiting factor - the eye may detect a single photon, but you are unlikely to
perceive this weak signal - after all, such a weak signal is hard to distinguish from electrical noise in the
optic nerve. The same is true for robots - designing robots with superhuman senses is not as easy as one
might think! First you need a lot of computing power, and second you need a high signal to noise ratio.
Cooling electronics and the use of superconductors and optic fibres may reduce noise, but the laws of
physics mean that noise can never be reduced to zero. Ever been highly alert, listening for a tiny sound,
and have you then noticed how false alarms increase in frequency - did you hear someone's shoe squeek,
or did you imagine it? That is because when you are on full alert, the threshold at which the brain ignores
noise is lowered so that the senses become more sensitive, but at the increased risk of false signals due
to noise in the system. You couldn't concentrate like that all the time though, your brain would tire from
working too hard. Likewise robots can be given highly sensitive sensors with variable thresholds for cutting
out noise, but practical limits to sensitivity will always exist.
Think also how much the brain must process in order to make sense of what the eyes see. When you look
at a chair under a desk, you may not see the whole chair, but you will imagine its shape and know that it
(probably) has four legs. This is harder than it sounds. Imagine a computer looking at a simple image, how
would it decide where the chair ended and the desk began? First of all you need stereoscopic vision or
depth perception. This is made possible by having two eyes slightly spaced apart on your head - each
takes an image from a slightly different angle and by comparing these two images the brain can
reconstruct a three dimensional image complete with an accurate distance scale. Without this ability you
would find it harder to pick up a cup without missing it. Shadows, textures, movements and contrast also
gives additional clues. Similarly having two ears on either side of your head enables you to tell which
direction sound is coming from (off to the left or to the right) by measuring the time of arrival and intensity
of sound in both ears. If in doubt you will turn your head or move about a bit to get more spatial information.
9.Economics
In the end, even if you can design strong muscles, fast brains and all the rest, there will be a limit based,
not just on engineering trade-offs and compromises (such as the need to minimalise weight or the need to
sacrifice strength for speed) but also on cost, simply in terms of raw materials and the cost of obtaining
them. Perhaps, instead of a super-tough heavily armoured and heavily armed smart robot, half a dozen
simpler and cheaper units will be more effective. Often there is no simple answer. Living organisms face
the same problem - raw materials (such as proteins and calcium) and energy are not in limitless supply for
living organisms. Some creatures adopt the strategy of investing heavily in the individual (such as the
elephant or oak tree) in the hope that most individuals will be tough enough to last long enough to
reproduce. Alternatively, one could be small and breed fast (such as a rat or birch tree) and hope that by
producing so many individuals, enough will make it to reproduce, even if most do not. Humans were
originally somewhere in-between, and indeed infant mortalities were historically very high and still are in
many parts of the World. However, in more developed countries people have fewer children and invest
many more resources in each individual, which is more like the elephant's strategy. Economics will always
induce relative weaknesses in any machine - the armour of a battle tank is always thickest where it is most
likely to be hit, leaving it with a relatively weak rear and underbelly.
10. Fitness for purpose
In the end it all boils down to how well a machine performs the job it was built for. The Cybex 7000 has
already proven its worth many times in battle. The proof of the pudding is in the eating of it, though ever
more realistic virtual simulations help try and test robots. As for ASIMO - well, I find it hard to see what his
purpose is quite frankly, though he is an important development which will lead onto more potential
applications as the technology develops. ASIMO is being improved and upgraded all the time, so
congratulations Earthlings - you have truly joined the Age of Robots.
What is the ultimate machine!
There isn't one! A basic engineering principle says that all machines can be improved, at least in principle,
since no machine is 100% efficient or maximally powerful. It also depends what you want the machine to
do. Robotics is tough when it comes to manufacturing sophisticated weapons like the Cybex 7000 which
can operate independently for long periods of time. Already, simple machines can better humans at very
simple tasks for which the machines are specifically designed - humans were not designed to add lots of
numbers very quickly or to manufacture thousands of silicon chips! In the end, even when machines
outperform humans, humans developed, through evolution, on a planet containing nothing more than
natural resources. For robots to equal this characteristic strength of living things, they will need to multiply
themselves in a sustainable manner. Living organisms are not perfected, since evolution does not actually
perfect anything - an organism only needs to be able to compete against its current rivals. Designing
robots or even living organisms will enables development of machines better suited to specific tasks for
which they are designed, though engineers on Earth have a long way to go!
Click here to read an illustrated essay on artificial intelligence. This essay compare robot brains
(computers) with the human brain and addresses the question: can robots or computers be intelligent?
Cyberprime and Borgtech Corporations, click to enlarge. This is one of the most advanced weapon
systems of its kind in the known Universe. It has extreme mobility, and is capable of running at sustained
speeds of 100 kilometres per hour. Powered by an EPA antimatter 8X series generator the Cybex can
metres tall, the Cybex 7000 is extremely agile and its excellent fire control systems ensure near 100%
accuracy at ranges up to one kilometre. The actuators are operated by advanced polymeric
syntho-muscle fibres which give the Cybex the strength (though not the inertia) of some 20 men.
The Cybex 7000 has a quantum positronium CPU (QX 7990) which is only some 9 cm in diameter but is
capable of advanced and rapid parallel computations for battle-field situations. Some components of this
system require freezing to -100 degrees Celsius for optimum performance and the antimatter generator
can also generate large amounts of heat. In addition rapid movement and use of energy weapons all
generate surplus heat which must be dissipated, however, it is important to maintain a low heat signature
for stealth. To achieve this the Cybex 7000 has a thermal screen which converts heat and infrared
radiation into higher frequency electromagnetic energy, either optical light or ultraviolet. This light is
radiated through a crypsis screen which can match the outgoing light to the environment, and thereby
providing cryptic camouflage at the same time as radiating away excess heat energy whilst maintaining a
faint infrared signature. The crypsis screen can also absorb, scatter or reflect hostile laser beams, and
other heat attacks, minimising damage from laser hits.
Other defensive systems include a negative screen which provides protection against electrical attacks
and repels anyone who attempts to grapple the Cybex 7000 by administering a severe electrical shock on
contact. The microarchitecture of the armour system is designed to channel frequencies of ultrasound
commonly used in sonic beam weapons around the Cybex, so that the ultrasound passes without causing
damage. The armour itself is made up of tough composite materials to absorb kinetic energy impacts,
including a layer which dissipates shock waves that travel through the armour, and thereby protecting the
delicate internal components.
The Cybex 7000 can be equipped with a variety of weapons and tools, the one shown here is armed with 4
Androm X-9 particle cannons (PCs) on its head turret, which can alternate between pulsed or beam mode
and between negatively and positively charged particles. These cannons have a maximum effective range
of 500 metres. The right arm is carrying an Omega-5 Mk 6 variable wavelength armour-piercing laser
cannon (APLC), with an effective maximum range of 3 kilometres. These weapons can also be operated in
laser lance mode for drilling through especially tough armour plate, such as on the hull of a warship. The
left arm is carrying the Battletex Epsilon-theta Mk 5 heavy ion cannon (HIC) which can operate in beam
mode or in wide-field mode when used to disrupt communications and electrical equipment.
The principles of robotics
We shall compare robotic systems to living systems with particular reference to what is possibly the most
sophisticated robot so far developed by Earthlings, ASIMO developed by Honda, and the human being.
Search for ASIMO in Google or on Howstuffworks.com.
1. Power
Providing energy for robots is no small task. The average human male uses about 120 watts of power
averaged over the course of a day, with peak outputs reaching around 1000 W during sprinting. As a
comparison, 120 W is the typical power provided by a one square metre solar panel, and the power of the
Sun's rays striking the Earth's atmosphere is 1366 W per square metre. The most advanced robots built
by Earthlings to date, include the Honda ASIMO humanoid robot is powered by a 51.8 V lithium-ion battery
weighing 6 kilos, which provides enough energy for one hour before it needs recharging and ASIMO has
only the fraction of the athletic power of a real human.
2. Strength
ASIMO can only lift a few kilos in weight. However, human-sized androids have been developed in Japan
that can allegedly lift around 70 Kg. You might think that it was an easy task to give a robot strength, after
all heavy hydraulic machinery is enormously strong, however these systems do not scale down well to the
size of a human being. ASIMO uses 34 servo motors, or electric motors that work by rotating shafts. Such
motors are clearly power hungry. Human muscles contains arrays of protein filaments (polymers, molecular
motors) that slide together using chemical energy, causing the muscle to shorten (animal muscles always
pull and never push). Similarly, robot muscles may operate using polymer fibres that change their
conformation or shape, perhaps by coiling into tight helices when an electric current is applied, similarly
causing the muscle fibres to shorten. However, making a 100 kg robot that could lift 250 kg (and thus
match some of the strongest humans) is no easy task.
3. Agility
ASIMO movements are limited. Each of its joints is operated by an electric servo motor, of which there are
34 in all. Humans, in contrast, the human body has between 565 and 850 separate muscles (the exact
number varies between individual and some muscles may merge and so appear as a single muscle). This
gives the human body a much greater range of motion. However, powering 800 electric motors would
require considerably better batteries! Again, molecular motors may be a better solution. However, ASIMO
can walk down steps, run in a circle and lift objects without breaking them.
4. Speed
ASIMO can run at 6 kilometres per hour (compared to a jogging speed of about 13 kilometres per hour for
a reasonably fit human). In humans, the thickness of the muscle as well as the muscle type determines the
muscle's strength. However, longer muscles and longer limbs are faster, all else being equal. This is why
sprinters tend to be tall and javelin throwers tend to be tall to give their long arms speed. However, shorter
limbs are stronger, since they have better leverage, so we would have short and thick limbs for strength
and long limbs for speed. Power is another parameter, and is equal to the rate at which the muscles can
do work (the maximum rate at which they can expend useful energy) and this is proportional to the total
volume of the muscle, so for maximum power we would have long and thick limbs. This is one reason why
athletes from different sports are so different in body shape, especially at the elite end of the spectrum.
Muscles also come in two main types - one that is fast and strong but fatigues quickly, and so is useful for
sprinting, and one which is slower and weaker but fatigues much more slowly, and so is good for distance
running. The proportions of each muscle type that a human has depends largely upon genetics - some
people innately make better sprinters or weight-lifters whilst others make better marathon runners. In the
end it is impossible to maximise speed, strength, power and speed - at some point there is a trade-off and
one has to compromise or invest most heavily in whichever is most important for the specific purpose.
5. Weight
Humans are often conditioned into thinking that the heavier the better - since weight is good for pushing
smaller creatures out of the way. However, a larger body requires more energy to move and maintain and
more raw materials. If you look at a range of animal species working in Earth's gravity, then you will find
that the fastest animals tend to be about 100 kg in weight and smaller and larger animals tend to be slower
(though there are exceptions of course). Smaller animals lack the muscle power whilst larger animals
simply have too much bulk to move (though an elephant can still outrun a human over short
distances)Humans are actually rather slow animals with the fastest humans ranking mediocre in sprinting
speed. However, humans make very good distance runners (with training!). Having less weight to carry
certainly means that you use less energy when travelling over a greater distance. Constructing robots from
materials which are light and strong is a major goal. Lighter materials will mean that less power is needed
to move the robot. Sometimes weight is an advantage - when it comes to crashing through concrete walls,
for example, and some warbots are heavily built for this reason, just like heavy tanks. However, even in
main battle tanks, weight reduction is an issue - lighter armour means that more weight can be carried in
munitions, for example. Animals (and plants) are made out of materials that are enormously strong but
very light. Bone is stronger than steel reinforced concrete, but is a fraction the weight. The cellulose and
collagen that make up plant tissues are strong as or stronger than steel, but a fraction of the weight.
Carbon-based materials (including animal and plant tissue components) make very strong structures, such
as silk protein, but are much lighter than steel. Proteins have a consistency rather like plastic, but can be
much stronger. Spider silk has a strength comparable to synthetic carbon fibres or kevlar, but can be
made by the spider at room temperature and pressure, using natural and renewable resources.
6. Mental processes
So far the hardest task for robotics engineers on earth has been producing robot brains. ASIMO carries
two laptop computers in its backpack, but still requires a human controller with a joystick, although it
performs the details of its instructed tasks automatically, but it cannot think! Incorporating powerful
computers will also increase the power demands of a robot. The human brain has a power output of about
20 watts - it weighs less than 2% of the body's total weight, but consumes some 20% of the body's energy.
Current estimates place the processing power of the human brain at between 10^15 and 10^20 operations
per second, though there are reasons to expect that this could be 1000 times as great as this. This
compares to the Blue Gene supercomputer, the fastest on Earth with 3.6^14 operations per second
(strictly floating point operations per second, or flops per second). Such computers can simulate a
simplified mouse brain at one tenth normal speed, suggesting that the human brain's processing speed is
certainly nearer the upper end of the estimates given (say 10^16 operations per second or higher). The
Blue Gene has 2^16 compute nodes and some 64 cabinets. This compares with the human brain with
10^11 neurones (brain cells), each with some 1500 synapses on average (though some neurones may
have 10 000 synapses or so) gives the brain 10^14 to 10^15 compute nodes, but each is capable of
performing 10 operations per second (and possibly up to 1000), enabling the brain to perform say 10^16
operations per second. However, in addition to these neurones, there are ten to fifty times as many
neuroglial cells in the brain, some of which appear to have (unknown) functions in processing, enabling the
brain to perform perhaps 10^19 operations per second. There are other possibilities that may further
increase this number 10-100 hundred times. Thus, the Blue Gene supercomputer is probably some
100-1000 000 times short and it would take 100 Blue Genes to stand any chance at all of matching the
human brain and that's at least 6 400 cabinets consuming massive amounts of electricity (26 800 watts per
cabinet or 1.7 million watts in total!) compared to a 1.5 lb brain consuming 20 W! Nevertheless, it does
look within the grasp of Earthling technology to match this complexity soon. A typical desktop computer is
way behind at around 10^9 operations per second.
Quantum computers (such as that used by the Cybex 7000) have the potential to thrash the Blue Gene
supercomputer and use much less power and occupy much less space, but Earthlings will have to wait and
see how effective these quantum computers can be. At this point you might be thinking why is it then that
the Blue Gene can perform computations vastly more complex than what you can do in your head. The
answer is simply that most of the brain's function is automatic and subconscious and so you do not have
much control over it. That isn't to say that the brain is reckless, rather most of the brain is involved in
calculations concerning internal organs and body maintenance. Also when you catch a thrown ball, you do
not consciously compute velocities, trajectories and angles in order to compute your own trajectory to
intercept the ball - this is done for you, automatically by the vast computational power of your brain.
Programming ASIMO to do the same tasks is proving immensely difficult! Even the tiny brain of an insect
has some one million neurones and so may perform an estimated 10^11 to 10^16 operations per second,
still equivalent to a thousand or so desktop PCs!
7. Intelligent materials
Living organisms are made from intelligent materials, by which i mean materials that respond to
environmental stress, modifying themselves when they need to improve and repairing themselves when
damaged. Robots too can be constructed from intelligent materials. Sensors can detect the stresses and
strains placed on joints, limbs and motors and skin sensors can detect damage. Pain is a necessity for
conscious organisms. If you did not feel pain then you would not take all the necessary steps to avoid
damage. Some people are born without the genetic ability to perceive pain and these people die young. In
this way, pain is a useful survival aid (though like all systems it can go wrong in the case of useless and
debilitating chronic pain). Pain is often suppressed when an organism is in imminent danger - there are
many reports of soldiers, for instance, not feeling their wounds until help had arrived. Even a tough warbot
must be programmed to avoid damage (indeed warbots especially so) but we do not want inappropriate or
debilitating responses. If damaged, materials can be designed to repair themselves. For example, liquid
chemical constituents can be pumped to the damaged area where they can solidify to form a protective
scar, or if more advanced, they may self-assemble into new materials of the right type. Tiny robots
(possibly nanobots) inside the main robot can also assist with repairs, though there are other conceivable
methods. Think how short lived most of the machines around you are (cars and computers etc.) that is
partly because they are unable to repair even daily wear and tear. It is also because biological materials
are very durable - even if your bones could not repair (and they certainly do!) they would last some 20
years of normal usage. (Furthermore, bones actually get stronger with increased usage). The brain can
also respond with damage and wear by re-routing its circuits to make maximum use of available brain
tissue. If your desktop blows its CPU, then it is defunct. The Cybex 7000 incorporates repair and
maintenance systems and can remain active for decades, perhaps even centuries. The Cybex 7000 also
has the ability to reroute damaged circuitry in its central processor, so that quite extensive damage is
required to completely put it out of operation.
8. Sensors
If you took the eyes of an eagle, the nose of an elephant, the ears of a bat (for sake of arguments) and
the ability of a goldfish to see in infrared and ultraviolet as well as in colour, and the sense of direction of a
homing-pigeon and you would have a very sensitive animal indeed! However, sensors are only as good as
the computer that processes the information they receive. Processing information received from the eyes
of a human being take up a large part of the brain, whilst the sense of smell takes up only a small part in
humans (but a large part in dogs) so to have all senses highly tuned would result in an excessively large
brain! Nevertheless, animals are extremely sensitive - their eyes can detect a single photon of light.
Background noise is another limiting factor - the eye may detect a single photon, but you are unlikely to
perceive this weak signal - after all, such a weak signal is hard to distinguish from electrical noise in the
optic nerve. The same is true for robots - designing robots with superhuman senses is not as easy as one
might think! First you need a lot of computing power, and second you need a high signal to noise ratio.
Cooling electronics and the use of superconductors and optic fibres may reduce noise, but the laws of
physics mean that noise can never be reduced to zero. Ever been highly alert, listening for a tiny sound,
and have you then noticed how false alarms increase in frequency - did you hear someone's shoe squeek,
or did you imagine it? That is because when you are on full alert, the threshold at which the brain ignores
noise is lowered so that the senses become more sensitive, but at the increased risk of false signals due
to noise in the system. You couldn't concentrate like that all the time though, your brain would tire from
working too hard. Likewise robots can be given highly sensitive sensors with variable thresholds for cutting
out noise, but practical limits to sensitivity will always exist.
Think also how much the brain must process in order to make sense of what the eyes see. When you look
at a chair under a desk, you may not see the whole chair, but you will imagine its shape and know that it
(probably) has four legs. This is harder than it sounds. Imagine a computer looking at a simple image, how
would it decide where the chair ended and the desk began? First of all you need stereoscopic vision or
depth perception. This is made possible by having two eyes slightly spaced apart on your head - each
takes an image from a slightly different angle and by comparing these two images the brain can
reconstruct a three dimensional image complete with an accurate distance scale. Without this ability you
would find it harder to pick up a cup without missing it. Shadows, textures, movements and contrast also
gives additional clues. Similarly having two ears on either side of your head enables you to tell which
direction sound is coming from (off to the left or to the right) by measuring the time of arrival and intensity
of sound in both ears. If in doubt you will turn your head or move about a bit to get more spatial information.
9.Economics
In the end, even if you can design strong muscles, fast brains and all the rest, there will be a limit based,
not just on engineering trade-offs and compromises (such as the need to minimalise weight or the need to
sacrifice strength for speed) but also on cost, simply in terms of raw materials and the cost of obtaining
them. Perhaps, instead of a super-tough heavily armoured and heavily armed smart robot, half a dozen
simpler and cheaper units will be more effective. Often there is no simple answer. Living organisms face
the same problem - raw materials (such as proteins and calcium) and energy are not in limitless supply for
living organisms. Some creatures adopt the strategy of investing heavily in the individual (such as the
elephant or oak tree) in the hope that most individuals will be tough enough to last long enough to
reproduce. Alternatively, one could be small and breed fast (such as a rat or birch tree) and hope that by
producing so many individuals, enough will make it to reproduce, even if most do not. Humans were
originally somewhere in-between, and indeed infant mortalities were historically very high and still are in
many parts of the World. However, in more developed countries people have fewer children and invest
many more resources in each individual, which is more like the elephant's strategy. Economics will always
induce relative weaknesses in any machine - the armour of a battle tank is always thickest where it is most
likely to be hit, leaving it with a relatively weak rear and underbelly.
10. Fitness for purpose
In the end it all boils down to how well a machine performs the job it was built for. The Cybex 7000 has
already proven its worth many times in battle. The proof of the pudding is in the eating of it, though ever
more realistic virtual simulations help try and test robots. As for ASIMO - well, I find it hard to see what his
purpose is quite frankly, though he is an important development which will lead onto more potential
applications as the technology develops. ASIMO is being improved and upgraded all the time, so
congratulations Earthlings - you have truly joined the Age of Robots.
What is the ultimate machine!
There isn't one! A basic engineering principle says that all machines can be improved, at least in principle,
since no machine is 100% efficient or maximally powerful. It also depends what you want the machine to
do. Robotics is tough when it comes to manufacturing sophisticated weapons like the Cybex 7000 which
can operate independently for long periods of time. Already, simple machines can better humans at very
simple tasks for which the machines are specifically designed - humans were not designed to add lots of
numbers very quickly or to manufacture thousands of silicon chips! In the end, even when machines
outperform humans, humans developed, through evolution, on a planet containing nothing more than
natural resources. For robots to equal this characteristic strength of living things, they will need to multiply
themselves in a sustainable manner. Living organisms are not perfected, since evolution does not actually
perfect anything - an organism only needs to be able to compete against its current rivals. Designing
robots or even living organisms will enables development of machines better suited to specific tasks for
which they are designed, though engineers on Earth have a long way to go!
Click here to read an illustrated essay on artificial intelligence. This essay compare robot brains
(computers) with the human brain and addresses the question: can robots or computers be intelligent?
Click here to see how robots can be
used in space exploration!
used in space exploration!
 |  |  |  |  |  |