Robotics get to the heart of the matter

Babbage recently underwent a small operation to remove a benign intestinal polyp, and was most impressed by the surgical technology. He was not etherized, but merely under conscious sedation. Nor was any incision involved; the procedure was conducted entirely via a prehensile probe equipped with a fibre-optic camera and biopsy tools.

For more major operations too, the safety advantages of minimally invasive procedures are becoming well-known. In traditional heart surgery, a striking development is the so-called 'daVinci' machine now in some medical facilities. Akin to the Waldo remote manipulator, a daVinci probe works, within the chest of a patient, as the surgeon's eyes and inhumanly steady hands. The surgeon can operate, without fear of 'the shakes', from a non-sterile environment, while patient trauma is greatly reduced by the small 'keyhole' incisions rather than the usual single, massive sternum breach.

At present, daVinci is the size of a room, and only a tool. But the future implications are even more revolutionary. In a university prototyping lab, Babbage was shown an automated version - a shoebox-sized ‘daVinciette' - learning surgery. Its patients were eggs of plastic gel, each with a caseless clockwork watch for a heart, the surrounding gel skin embedded with wire mesh and red ink bubbles. Some watches were deliberately damaged, and the daVinciette's task was to mend these without interfering with the obstacles in the skin.

As Babbage watched, the daVinciette took each egg, and proceeded to identify its watch, penetrate the gel skin for an examination, then attempt repair when necessary. At each stage, it could refer the case for human attention. Finally it wound the mechanism, then classified the patient as dead (the watch not ticking, or ink in the works), alive, or to be referred. The daVinciette was not the best of surgeons: around a third of its patients died, and one in ten was referred. Nevertheless, its record was slowly improving, and viewed alongside other medical robotics advances, this brings the 'Aesculapius machines' and 'autodocs' of science fiction into the realm of likely futures.

Of the medical endorobotics experts Babbage spoke to, some foresaw automated surgery useful in battlefield or other hazardous environments. Others saw brighter prospects, a way out of resource shortages in a world of escalating clinical demand. But whatever the applications for daVinci and the daVinciettes, their increasing part in medicine seems to Babbage inevitable.

Escaping from the footnotes of history

It is always pleasant to meet kindred spirits, and Babbage recalls his introduction in 1840 to fellow Devonian Mr Thomas Fowler. A bank manager from Torrington, Fowler was visiting London to demonstrate his wooden calculating engine. Babbage thought that the device, of sliding-rod construction and using the ternary number system, showed remarkable promise. Its inventor, alas, died three years later, leaving it in pieces. Soured by the experience of competitors copying his Thermosiphon central heating system, Mr Fowler left no drawings for the calculator’s reconstruction. The only record is a commemorative window in St Michael’s Church, Torrington.

Unfortunately this is all too common a story. In our electronic age, mechanical computers seem destined to be mere quaint footnotes to history. In reality, their significance was considerable. Notably, the Turing ‘Bombe’, crucial to Britain's decoding of German Enigma messages in World War II, was a programmable mechanical computer, albeit not “absolutely general” machine like the Analytical Engine.

Mr James Redin’s “X-number World of Calculators” at http://www.dotpoint.com/xnumber/vintage.htm celebrates hand-cranked comptometers, slide-adders, and other descendants of early devices such as the Pascaline and Leibniz Stepped Drum calculators. Mechanical calculators were the mainstay of businesses until only a few decades ago. One of the earliest such devices, the abacus, is still in use today.

Recent developments, however, may yet revive mechanical calculation in mainstream science. At the web site of Sandia National Laboratories, USA (http://www.mdl.sandia.gov/micromachine/) Babbage recently learned of some delightful ‘micromachines’ no larger than a pollen grain. Formed as an integral part of a CMOS microchip, these devices include locks, shutters, sensors, encoders, and even a steam engine.

Moving down to atomic scales takes us into the realm of Dr K Eric Drexler’s ‘nanotechnology’. Atomic-scale gears and bearings are theoretically possible, and from these, nanocomputers could be built. Babbage finds most amusing the thought of a nano-sized Analytical Engine driven by a steam motor little larger. A more likely component, however, is the sliding ‘molecular abacus’ reported in the February/March 1999 issue of Scientific Computing World. It seems that Mr Fowler’s sliding-rod concept, albeit on a far different scale from his prototype, could at last find application.

---

A footnote from Babbage: I was delighted to read in the Western Morning News, August 30th 2000, an item about two reconstructions of Mr Fowler's Engine by Roy Foster, an engineer from Torrington, and Mark Glusker, a product designer from California. The construction was made possible through the work of historian Pamela Vass, also of Torrington, whose research led to the rediscovery of Fowler's notes on his Engine's design, given as a deathbed dictation to his daughter. A short biography of Fowler may be found at www.thomasfowler.org.uk; Mark Glusker's model may be seen at www.mortati.com/glusker/.

Virtual recipes for a primaeval soup

The passing of a year stirs recollections; and Babbage recalls with regret his travails in communicating to successive governments the vast possibilities of the Analytical Engine. But he was recently privileged to attend a policy studies ‘think-tank’, where he was startled to hear sober scientists, industrialists, civil servants and politicians considering the scenario of self-aware intelligence arising, unplanned, within the structure of the Internet.

Surprisingly few delegates rejected the idea, though a unified ‘spirit of the Internet’ was dismissed as a fairy-tale concept. The rest tended to conclude that the present Internet closely resembles the primaeval biotic "soup", and rapidly assumed that such intelligence would be not singular but plural.

The exact nature of potential intelligences was hotly debated. Would there be a few powerful mentalities, perhaps even superhuman? Flocks of mutating "shades" of vertebrate-like intelligence? Or a myriad of individually mindless beings summing to a whole, like termites? Babbage was astonished to hear, despite disagreements, estimates that such entities could appear within seven to fifty years.

One delegate asked about the prospects for exterminating such entities were it deemed necessary. There was broad agreement that attempts were inevitable, yet pointless and even counter-productive. Starvation, the denial of computational resources, would damage infrastructures dependent on the Net. Engineering digital predators could accelerate the evolutionary process in the survivors if it failed. If successful, such predators could present a new, more virulent, problem of their own.

How worthwhile are such conclusions? On hearing Babbage’s account, Lovelace commented that some of the delegates’ assumptions seemed simple analogies of the problems of dealing with biological pests. This may be valid; at the 1997 Digital Burgess Conference (see http://www.biota.org/) biologists likened current digital biology to the multifarious organisms of the Pre-Cambrian Burgess Shale. But some assumptions seemed closer to images from popular culture: the fear of a hostile Internet recalls Terminator and many other cinematic visions.

Making his way back to his hotel, Babbage wondered how Lovelace's transplanted human souls and the predicted natives would react to each other - and how he came to be seriously asking himself this question. He had longed for recognition of the Analytical Engine in decision-making circles, but this gathering left him feeling he had imbibed a dose of Lovelace's laudanum.

From cyborgs to wearable computers

Babbage has, on occasion, expressed misgivings about the various interpenetrations of silicon and flesh. Why risk ‘cyborging’ oneself, when a wearable Analytical Engine can upgrade natural faculties through a removable carapace of information technology?

Babbage has encountered a number of such Engines, whose apparent variety simplifies to three essential types. First, and most straightforward, are those which deliver to their wearer a body of expertise. Babbage was recently shown helmet-mounted rigs for use by emergency and other workers in locations where expert support would be otherwise impossible. While video and audio devices provide outward communication, incoming data presentation derives from the ‘head-up displays’ of military aviators: half-silvered visors and mesh diaphragm earpieces that overlay vision and hearing.

A second type uses similar methods for controlling remote devices from a safer location. Babbage revisited a North Sea fish farm where he had been required to flounder in a diving suit through cold, murky waters. This time, he was given a headset and belt control pack, and instead promenaded around catwalks in pleasant sunshine. Some fathoms below, a small submersible vehicle kept automatic pace, relaying hydrophonic sounds and six video channels for Babbage’s edification.

A more alarming setting demonstrated the third type: that which amplifies the wearer's sensory analysis. In an urban warfare training ground, Babbage watched an infantry patrol under simulated sniper fire. Even as one man rolled for cover, a computer about his person was analysing the sound and its echoes, passing the source coordinates both to a wrist display and as aiming data to the optical sight of his weapon.

Although somewhat shaken, Babbage considered other applications. Sudden loud sounds might equally signal danger to firefighters or workers in volcanic areas, as could subtler sounds: the creak of a stressed component, or the rush of leaking industrial coolant. Other senses might be similarly augmented, and combination with headset displays could allow whole teams to track invisible and inaudible phenomena - say, a flock of bats in the dark - through a shared virtual environment.

This technology is closer to the everyday than Babbage had imagined. With many people already routinely carrying telecommunications and palmtop computing devices, and video cameras costing no more than a waterproof coat, no great conceptual leap is required to envision their connection. Who can say to what uses science might not put this sensory expansion?

Artificial life begins its own evolution

“To every thing there is a season,” says the Book of Ecclesiastes, and Babbage finds it satisfying that two scientific advances that ‘missed’ each other historically should at last find their season and flourish together. When Mr Charles Darwin set out his theory of evolution in The Origin of Species in 1859, Babbage’s computing engines were already a curiosity, unfunded and fading from official memory. But now, as a new millennium approaches, Darwinism and Analytical Engine are intricately entwined, and we are seeing Engine programs that evolve over time via a process of mutation, mating and selection.

A brief excursion upon the Internet finds many examples of genetic programming, which evolves populations of programs in a mimicry of Darwin's ‘survival of the fittest’. In the field of artificial intelligence (AI) especially, this solves problems ranging from the academic, such as the University of Michigan's GAIA agent which analyses radar images to classify sea ice roughness, to the recreational, such as the Spanish GENEURA team’s MasterMind game server.

Genetic programs are often teamed with neural networks, another tool developed from the same biological analogy. Examples of neural-genetic algorithms include Advanced Investment Technology's stocks and shares analyst, and the Danish-Italian art software, Artificial Painter.

However, a British company, CyberLife (http://www.cyberlife.co.uk) has gone further with an innovative marriage of AI with biochemistry. Its synthetic organisms, ‘norns’, possess neural network brains bathed in a sea of virtual biochemicals that stimulate or suppress their response to stimuli. In an ecology of virtual plants and animals, the norns adapt over generations through natural selection. Their present applications include a charming software toy, Creatures, but norns might see a darker future as artificial military pilots.

Are norns truly autonomous? Babbage recalls Lovelace’s view - that the Analytical Engine is limited to “whatever we know how to order it to perform” - as but one expression of the claim that AI is, at heart, merely a wealth of carefully encoded human experience. Even if not consciously applied, this experience might still colour the programmers’ assumptions of the rules underlying life.

But with norns, interestingly, unprogrammed properties have emerged, such as bees swarming, and plants adapting to different conditions. Babbage wonders if technology is poised to evolve in ways that take it beyond human understanding. Perhaps with these techniques, we are beginning to witness not just life as we know it, but life as it could be?

Cyborgs: bodies enhanced, or invaded?

The limitations of the body are frustrating, as Babbage found when a sore throat consequent upon an attack of influenza left him incapable even of railing at the buskers who congregate outside whenever he takes to his sickbed. Voiceless, he felt in a position similar to that of the unfortunate San Francisco gorilla, Coco, who is to be conscripted into using the Internet in the latest of a tradition of experiments in teaching human language to apes.

Babbage firmly believes that experiments of the Washoe and Nim Chimpsky ilk obscure more than they reveal about language, thought and intelligence. Nevertheless, the mechanics of the situation are relevant to us all, since Coco's access to the Analytical Engine is necessarily indirect: Coco signs manually to a human keyboard operator, who reads aloud the text from the screen. This two-mode approach to duplex communication applies equally to human-Engine interaction, so limited in its efficiency by manual dexterity and the available bandwidth of the Engine's visual and audible replies.

Science fiction has long explored means of improving this communication by direct neural interfacing, or 'cyborging'. In Mr Samuel R Delany's 1968 novel, Nova, a spaceship's crew routinely plug control lines into electronic wrist sockets. Mr DG Compton's 1974 novel, The Continuous Katherine Mortenhoe features a journalist sending video images from his electronically supplemented eyes; and the "Bionic Woman" of the 1976 television series had multiple prostheses including an amplified electronic ear.

After three decades, such ideas have moved closer to reality. Servomotors in prosthetic limbs may be controlled to some extent by impulses from the remaining motor nerves. A research team under Dr Wentai Liu of the North Carolina State University has prototypes for partial restoration of sight via an electronic 'artificial retina'. And cochlear implants, stimulating sensory nerves, help in some cases of hearing impairment.

The fictional cyborgs are still of varying plausibility. But a University of Tokyo team has already produced a cyborg cockroach, controlled by electrical pulses to its antennae stumps, and cyborgs of a higher order would have obvious industrial value. Direct neural interfacing, too, cannot be far in the future, following last year's Caltech breakthrough in devising the first 'neurochip' of rat neurones grown through a silicon array. It seems that Coco's successors may, before very long, find themselves accessing their Analytical Engines far more intimately than at present.

Lateral thoughts both large and small

It is Babbage’s view that any fact might ultimately be useful, whether it be the heart rate of a pig, the proportion of sexes amongst poultry, or the tabulated causes of breakage of plate glass windows. Trivia, some may think. But a wealth of random knowledge is the basis for serendipity - chance associations that lead to fruitful discoveries - and Babbage was recently musing on the nature of such happy accidents in relation to the Analytical Engine.

The problem-solving power of chance association, whether called "intuition" or "lateral thinking", has been a vital ingredient in our evolutionary success. Its value has been immortalised in mythology, from Herakles’ idea of diverting a river to clean the Augean stables, to the tales of Archimedes’ bath-time “Eureka!” and Isaac Newton’s apple. It is a prime example of a role where the human brain bests the Analytical Engine; yet logically it seems an area in which the Engine could supersede us in seeking unsuspected correlations between facts.

In theory, this needs three components: huge data stores; repeated trial-and-error associations between items; and rapid probability-based assessment of the worth of any trial. These, however, give contradictory requirements of size and speed. For conventional chips, the Moore’s Law model - an annual doubling of manufacturable circuit density - is already running into physical constraints. To achieve the necessary performance, the options are larger devices, or new miniaturisation techniques.

A truly colossal device is already with us: the Internet. But what is waiting to be born is a vast distributed Analytical Engine whose parts spontaneously compare and process data among themselves, just as social animals (humans included) share knowledge and decision-making across their group. Babbage dubs this “metamachine” the Gaia Engine, in whimsical homage to Mr James Lovelock.

At the other extreme, we see the portents of engineering on the microscopic scale of animal nervous systems: for instance, the neurology of the locust’s flight control is well understood as a servomechanism; and Israeli scientists have produced short lengths of silver-plated DNA wire. Who can doubt that logic gates, then large scale integration - the ‘biochips’ of Mr William Gibson’s novels - will follow?

With such developments combined, Babbage's informants predict, we may soon see the day when the Gaia Engine will stir, try out a hunch and intuition or two, then settle down to some truly serendipitous lateral thinking.

Automata swarm with good intentions

Babbage is often delighted by the way modern technology brings happy reality to ancient dreams. Seeing hang-gliders over the Devon cliffs reminds him of the artificial wings of Daedalus. The vacuum flask, wherein he keeps his tea on railway journeys, recalls the magical bottles of Welsh legend which preserved the heat of any liquid, “though it was carried from the east of the world to the west”. An informant at a Swiss-based food multinational suggests another example on the near horizon, arguing that a ‘swarm’ of small robots could replace large agricultural machinery, labouring like the ants of Greek myth who helped Psyche fulfil the impossible task of sorting a large heap of mixed seeds.

This hopeful vision was echoed by the correspondents who took Babbage to task for his pessimistic picture of distributed automata as a potential monster. Many specialists, to whom Babbage can only defer, have painted instead a bright future of robot swarms in the service of humanity, from exploring the vastness of space to maintenance of the human body (Lovelace’s “molecular universe”).

An acquaintance of Babbage pointed to the contemporary problem of landmine clearance. Could not connected automata, she asked, perform this task more rapidly and safely than humans? Workers at a Scandinavian university quickly dispelled Babbage’s doubts as to the cost: each automaton could further disperse by controlling cheaper, more expendable, ‘drones’ to spread its presence across perhaps a hectare of territory

From an eminent American astronomer came the thought that dispersal could be applied beyond land devices. Flying, seagoing or spaceborne swarms are equally feasible, perhaps with different drone types held as a common resource and assigned to individual ‘queens’ as needed. They could operate in hazardous environments - ocean trenches, volcanic regions or the asteroid belt - taking over roles such as seismic research, disaster relief and mining. Such a swarm could even undertake for us the long journey to other stars. When idle in transit it could disperse to form a vast radio telescope, thousands of miles across: the ultimately flexible research tool.

Babbage finds this flexibility enchanting: an extension of Lovelace’s idea of the Analytical Engine as an absolutely general machine. And with advances such as the PatMax improved machine vision and the University of York’s AURA “reasoning” neural network, both described in November’s issue, these dreamt-of automata come ever closer to reality.

The double-edged sword of autonomy

Babbage was recently intrigued to read The Man who was Frankenstein, in which Mr Peter Haining argues that Babbage and Lovelace’s mutual friend, the electrical experimenter Mr Andrew Crosse, was Mrs Mary Shelley’s chief inspiration for her fictional scientist. Babbage finds this plausible, remembering well the moral outcry at Crosse’s later and much-disputed creation of insects - Acarus mites - by the action of the voltaic battery.

After 150 years, however, the idea of artificial organisms has become less outrageous. Machines are gaining autonomy: for instance, the remotely controlled vehicle exploring the planet Mars can refuse instructions that might endanger its safety. Babbage envisaged the Analytical Engine as a tool merely for amplifying the powers of human thought, and even Lovelace, with her greater prescience, said it would have “no pretensions whatever to originate anything”. Both are amazed, however, by the progress toward apparently sentient Engines.

In a European military research establishment, Babbage was shown rabbit-sized roving automata, equipped with jointed manipulator arms and electronic ‘eyes’ and ‘ears’. When not under direct control, they randomly patrol their environs to learn its geography. To ensure their own survival, they hide from any large moving object, and seek out electrical sockets to ‘feed’ (Babbage is reminded of Crosse’s ‘Acarids’). The military staff casually mentioned a future capacity for lethal force, tempting Babbage into considering noise-sensing automata as a stern measure against raucous street music.

In the same establishment, staff carry portable Analytical Engines interconnected via a wireless telephony network. Could the small vehicles be similarly linked, Babbage wondered? Yes, he was assured -- very easily. It is not, he feels, a very great leap to imagine a linked ‘swarm’ of such vehicles, comprising, like ants, a greatly flexible, dispersed entity. The loss of a single automaton would not fatally injure the swarm, which could repair itself or even manufacture complete new vehicles. Perhaps such swarms are the future of planetary exploration and much else besides.

Human sentience, it has been argued, derives from manipulative capacity and a critical number of neural connections. How large must a swarm be, Babbage wonders, to provide that critical number? He finds the question fascinating; but with the prospect of automata free to decide to copy themselves or to kill, Mary Shelley’s warning about irresponsible scientific creation carries as much weight today as it did when Babbage, as a young man, first read Frankenstein.

From the data of Nature to the nature of data.

Babbage was impressed, during his recent travels, by the variety of ways in which the descendants of his Difference Engine are now applied to the study of Natural Science. He was particularly intrigued by the intertwining of Computer Science, once a construct whose ideal was Mr Boole's pure symbolic logic, with the more anarchic Life Sciences.

In Proverbs 6:6, the Bible tells us to "go to the ant ... and consider her ways". In Martlesham Heath, England, telecommunications engineers have taken that advice, devising programs that direct telephone calls using ant-like behaviour to seek the best route. Independent software entities - 'agents' - move about the network leaving guiding messages for each other, just as ants leave scent trails to guide others to food.

In Stuttgart, physical chemists use automatic sampling of the statistics on a factory's processes to tune, as it were, the variables controlling the factory. They are, in fact, inducing a machine to mimic the trial-and-improvement patterns whereby an animal solves a problem.

In Paris, the seat of the perfume industry, other chemists link computers to chemical sensors which complement or augment the olfactory discrimination of the human nose. In Moscow, a computer simulation explores the possible futures of an endlessly mutated DNA chain.

The phenomenon is not merely one of learning to copy natural methods. Biologist Mr Stephen Jay Gould has memorably described the cloud of pollen from a tree as "raining floppy disks", and this view of Nature as information (utterly alien to the scientific paradigm of Babbage's day) is reshaping the endeavours of humankind to emulate those of natural creation.

Confused and dazzled by such complexities, Babbage took refuge in observation of humble lichens within an ancient oak wood in the Devon of his boyhood. When, however, one of these enigmatic plants defied recognition, he sought enlightenment in an Antwerp database. In pursuit of the Norwegian Bryoria smithii, it occurred to him that the branching hierarchy of this botanical key replicated the evolutionary and taxonomic classifications it described. And beyond, the global web (the key being but a small part) has itself developed to become like an evolving organism, driven less by its makers than by Mr Darwin's laws.

Such esoteric encapsulations of human data and universal code, Babbage decided, were the territory of Lovelace's more freewheeling intellect, and he resolved to concentrate his mind on comparatively straightforward considerations of engineering and physical reality.