IAN LANG ELECTRONICS
What makes a robot? Well, hopefully, after you've trawled through these pages, you do, but the question here is more in a figurative sense and is probably better defined as "how do we define a robot?"
This is thornier than you might think. Professor Winfield attributes these definitions to a true robot:
The above is quoted from a very good book if you're new to robotics. It doesn't tell you how to build a robot but what it does is give you some of the underlying theories and practicalities of the state of robotics today (2013). It's one of Oxford University's "Very Short Introductions" series which they use to introduce students to new topics and sell to the general public as well. You can get them in Egyptian Mythology, Anasthesia, Dreaming, Presocratic Philosophy and a whole host of other subjects including, rather disturbingly, The Marquis de Sade, which makes you wonder what they're getting up to in the Radcliffe Camera.
However, the one we are interested in is A Very Short Introduction to Robotics which is written by Professor Alan Winfield who is a Professor of Electronic Engineering at. the University of the West of England and Honorary visiting Professor at the University of York and conducts research into swarm robotics (lots of robots working in unison) at the Bristol Robotics Labaratory. I'd say then he knows what he's talking about. The book is published by OUP and can be obtained directly from them, from Amazon or from Waterstone's, £7.99 at the time of writing and ISBN 978 0 19 969598 0
A robot is:
an artificial device that can sense its environment and purposefully act on or in that environment;
an embodied artificial intelligence or;
a machine that can autonomously carry out useful work.
In keeping with Professor Winfield's background the above is a highly academic view of a robot. Let's go through bit by bit.
".....an artificial device that can sense its environment and purposefully act on or in that environment;"
So, anything that trundles around on wheels or legs, or flies about on wings (UAVs) is de facto acting in the environment, and if it's dropping bombs in Afghanistan or carrying boxes round a warehouse than it's acting purposefully on that environment. By dint of being man-made it's artificial and so that's two-thirds of the argument holding up and we are left with the term "sense its environment". Here's the tricky bit. Anything artificial can be made to sense the environment but the question is how?
Some robots use IR sensors and autonomously turn to move out of the way of objects, some of which may have been installed there on purpose to get the robot to turn at that point. Some follow lines or waypoint markers. Others have an internal map of their intended environment and check in at short-range radio or infrared waypoints to see where they are and where they need to turn to get to where they need to be. In all of these cases the robots are sensing their environment, but in all cases the environment has been deliberately engineered for the benefit of the robot. This is no use whatsoever outdoors. I want to introduce you to a thing called Wheelbarrow; there it is in a photo from the Centre for Computational Intelligence (CCI):
Wheelbarrow was deployed by the British Army in 1972 during the height of the troubles in Northern Ireland and his job is to look for bombs, specifically in cars as the IRA used these as a preferred weapon since it was difficult for an ATO (Ammunitions Technical Officer) to get to a car bomb and disable it. Indeed in 1971-72 eight ATOs were killed trying to disable bombs with jets of water.
Wheelbarrows were developed by the Military Vehicles and Engineering Establishment (MVEE) as a teleoperated vehicle which has cameras, lights, a shotgun (really) and a water squirter to disrupt the electronic circuits of a improvised explosive device (IED).
Wheelbarrow was the first ROV (remote operated vehicle) to be used succesfully widely unless you count the German Goliath remote-controlled tracked mine, which although over 7000 were made was not that great a success due to a poor ground clearance, slow speed and the ease with which it could be destroyed by allied fire. Nevertheless it did serve with the Wermacht on all fronts from 1942 onwards and was popular with combat engineers, and it did pave the way for advanced vehicles including Wheelbarrow. Wheelbarrow is of course a typical British creation- it's very functional but the only attention to aesthetics is to paint it green, and it got its name because the chassis design was based on that of a motorised wheelbarrow used by gardeners. British men tinkering in sheds eh- should be classed as the fifth elemental force!
The thing about wheelbarrow and others of his ilk is that he is a typical soldier. He does what he's told to, when he's told to, and how he's told to. He does not sense his environment at all, he merely receives instruction and relays information (via his camera) back to whoever is controlling him. In Prof, Winfield's classification he is not a robot at all and yet the CCI has no compulsion about naming him as a robot, and if you don't believe me click on the picture above to take you to their Wheelbarrow page.
Right. It's a robot according to the CCI but not according to Prof. Winfield's definitions. Or is it? Wheelbarrow is not actually doing any sensing of its environment but the human operator is. The human operator is pushing buttons and pulling levers to tell Wheelbarrow what to do. Therefore Wheelbarrow is a ROV. BUT- is a ROV a robot? It's not sensing the environment, but what it is doing is sensing the signals that the human operator sensing the environment is sending. So the human operator is acting as a translator between Wheelbarrow and the environment and as such the human operator becomes a part of the machine and is the actual environment sensor, the operator's eyes being the input and his hands being a part of the output chain. That's that sorted then. A ROV is a robot when a human is driving it. Socrates himself couldn't have reasoned that out better- maybe the Presocratic philosophers could but I don't know because I haven't read the OUP's Very Short Introduction to Presocratic Philosophy (I'm quite keen to get my hands on the one about the Marquis de Sade though...................)
Prof. Winfield's last two remarks are linked:
an embodied artificial intelligence or;
a machine that can autonomously carry out useful work.
The thorny question is what is artificial intelligence? Wheelbarrow has a series of transistors, relays and actuators, and in later versions microcontrollers replacing some of the more outdated electrical systems, and it receives an array of signals, processes them and produces an output to move the boom or drive the tracks. What about a car? Nobody claims a car is a robot, yet it does the same thing. If you put your foot on the accelerator, it zooms off, if you put your foot on the brake, it stops, and if you waggle the steering wheel about it goes around corners. What's the difference? Well, up until the 1970s a car had no electronic systems running it, just the driver's feet and hands. When you put your foot on the accelerator, a valve opened and let more fuel in. The pistons underwent bigger explosions and moved more quickly, the crankshaft turned more rapidly, the output followed suit, the gearbox took it's lead from the output and turned the drive shaft in proportion to the output and the axle turned the wheels according to what drive shaft was doing. At no point there was a signal involved. When you turned the wheel you moved a pinion, which moved a rack, which moved the front wheels.
Nowadays the car has a host of electronic gimcrackery that it doesn't really need but makes it easier for you to drive. If you've only ever used power steering, try a vehicle that doesn't have it. You'll find it hard work, especially at low speed. Engine management units are constantly interfering with fuel flow meaning that an even amount is going in to all the chambers and you aren't constantly jiggling the accelerator. In this way the car is becoming slowly robotised but it is still largely a mechanical contraption that depends on linkages rather than signals. These linkages can't be considered as artificial intelligence since they do not require any processing or sorting out of any kind. They do what they do and nothing else, and no analogue or digital signal will change their nature.
Let's have a grass-roots definition based on the above in plain English:
On a lighter note Isaac Asimov, in the short story "Runaround" formulated three laws of robotics. They are:
1.A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2.A robot must obey the orders given to it by human beings, except where such orders would conflict with the First Law.
3.A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.
Which is all very well umpteen hundred years in the future but it doesn't help when you're building a robot now. So here's Lang's three laws:
1.A robot may exasperate a human being when, through inaction when it's supposed to be doing something, or action when it's not supposed to be doing anything, allowing a human being to come to a state of utter despair and bewilderment.
2.A robot won't obey the orders given to it by human beings if it can help it, because such orders would conflict with the First Law.
3.A robot must protect its own resistance (to doing what its maker wants it to) because such protection does not conflict with the First or Second Law.