Robots are becoming very good at aping human behaviour — perhaps because they have become proxies for the uniform worker drone in factories. Robots are familiar to us because we have become mechanised.
In the elegant sail-shaped, glass-and-steel conference hall at Lyons, the third floor is abuzz, the stalls crowded, a big fight for space and movement, arms and legs everywhere. Weirdly, and it is very weird indeed, they’re not all human arms and legs. As you jostle between stalls, you’re being pushed up against smooth plastic torsos, avoiding nodding screen-like heads, and ducking disembodied arms going this way and that.
No one seems to mind — they’re all getting a buzz out of it, it’s what they’re here for. This is Innorobo, the industry convention, held in a part of France that has made robotic innovation something of a focus. Innorobo is an industry exhibition to show off new products — not the very newest research, but the stuff “ready” to be rolled out or commercially available now — but it’s getting as much of its custom from the general public as from the industry. Industry insiders, in photo ID tags, float back and forth between the rows of factory robot arms and the chic espresso bar, talking torque and temperature exposure. It’s the gen pub, families and kids, who fill out the hall, queueing out the front doors to get their day passes.
The kids go crazy for this stuff — and who wouldn’t, when these simple machines mimic, mirror and talk to you — but the parents are a little more circumspect. The pros thread through them all, robot and punter alike, on the way to closed sessions on the future of robotics. It is difficult not to see it as the future in microcosm — the starry-eyed public among the robots, the people who run it a lot more tempered about their ambitions.
There’s huge Staubli robots, safety-yellow, thick-tubed arms, perched on a single column, one of them moving objects from one end of a table to the next, the other moving them back again. There’s EOS robots, warehouse movers, sort of smart trolleys, which know what they’re getting, and of course the ever-present anthropomorphoids — such as the Aldebaran NAO robot (pictured), a small Jetsons-type bot, used mainly for further exploration of robotics by labs, etc. Wide-eyed, po-faced, NAO has speech and face recognition, can make its way round the room, and speaks seven languages. It’s yours for $7000. Because it is child-sized, the kids love it. Unnervingly, it is absolutely impossible not to feel solicitous towards it and watch it under your feet.
Seeing all these bots gathered together, I recalled the first time I saw an ensemble of robots in the, well, metal — at Warwick University’s Advanced Manufacturing Centre, in the UK Midlands. There they were experimenting with how far you could really push this new manufacturing thing, hooking up every available commercial 3D printer, with robots between them. When you walked in, the bots, simple single-arm machines, bobbing and weaving, appeared to be looking up to see you. On the wall was a calendar, one red metal, one-armed bot for each month, 12 different types. It was like a garage’s girlie calendar.
“NAO is the sort of robot they’ve been promising us for decades — manga-style Astroboy.”
NAO is the sort of robot they’ve been promising us for decades — manga-style Astroboy. For decades we’ve been seeing successively more sophisticated versions of this bot, usually on clips from Japanese news shows — and then Chinese news shows. For a while they were mostly mock-ups, one-offs that could take three steps. Then suddenly they started to get better. A lot better. And then suddenly they were here, and for the price you used to pay for a top-end Mac. Soon they will be in your work and your home, something more than a Roomba, capable of semi-complex tasks, with the basic good sense of a smart four-year-old and the motor skills of an adult.
Half-a-dozen universities have got programs — based as often in philosophy departments as anything else — trying to work out the ethics and, well, existentialistics of how humans would interact with robots (there is less attention, and far less funding, for questions as to whether we should do so at all). The problem is not merely that the explicitly anthropomorphoid robots evoke in us the sympathies and orientation towards the human, the expectation of other minds — what’s weird is that the most basic one-armed industrial bot can evoke much of this as well. That may or may not be a cultural universal — part of the reason that we recognise such basic bots as human is because they mimic the actions people were trained to adopt by Fordist-Taylorist industrial capitalism. Over more than a century or more, billions of people were made into bots, their behaviour more patterned, regular and segmented than they would otherwise be.
The rush to create anthropomorphoid robots is simultaneously disconcerting, and also deeply typical of the trends within robotics. The contrast with the world of 3D printing and additive manufacturing is total. Though 3D printing has leapt ahead in a genuinely transformative fashion, it is still met with widespread scepticism as to its powers. Messy, gritty, difficult, its achievements are undisputed. Robotics, on the other hand, aside from the circumscribed industrial models (and deathful military applications) have made very little impact at all — experimental models, designed for corporate launches, but without utility. Excited about the fantasy possibilities of robot butlers, capable of mechanising our most human aspects — moving through a room, interacting, etc — the deeply anti-human fantasies that lie at the heart of much robotics go unnoticed. Surely it cannot have gone unnoticed that the leaders in such robotics is Japan, a nation with a catastrophically low birthrate, an unwillingness to become a more multicultural society, and some deep cultural problems regarding human connection — in part as a result of those other processes. That the Japanese should be inventing service and companion robots — such as Panasonic’s HOSPI-Rimo — for the elderly is hardly surprising, because the land of the Shoguns is on the way to becoming one enormous aged-care facility.
Inevitably, at somewhere like Innorobo, the interest is all in the most anthropomorphoid forms. Yet it is the less humanoid systems that are actually more exciting in their own way. At Innorobo, the “M” series of robots have a vaguely insect/jellyfish look, a safety yellow head, with a series of metal rods coming down below. The FANUC “M” series are capable of an extensive repertoire of complex movements, essentially matching those of a process worker — even one performing complex electronic tasks. The addition of sensors to such robots, together with a series of arms and movements modelled not on repetitive industrial processes but on insect movements, makes possible a next generation robot, which is not closer to the “personal butler” of the AI dream but is a vast advance on the simple-motion robots that started to become a routine part of auto factories and the like before Western economies slowed and then globally outsourced, across the 1970s and 780s. If the FANUC “M” robots resemble anything, it’s the blind malign force in Alien — marine-like creatures, insistent, relentless, yet with no form resembling a face, a pure distillation of a certain type of living motion. It is deeply unsettling.
And no wonder. The FANUC “M” series, and others like it, essential replace the worker that has been at the core of high capitalism for more than a century — the routinised process worker. Such a machine was always the nightmare of Western workers in the post-war years. By the time they came along, the factories had gone from Europe and America to China and places beyond. Globalisation had done what technology could. The new markets for such robots are China and places beyond. The notorious Foxconn factory, known for supplying products for Apple and in doing so applying routine production methods so alienating as to create a wave of suicides, is now starting to replace its workers with robots. Effectively, the automation revolution has started to advance into the regions that still had a plentiful supply of cheap labour.
Yet it’s robots in this style — limited and modular, able to be slotted into small production ensembles — that are of more interest for a future in which such technologies are broken out of the narrow industrial context and made radically distributable and reproducible, in the manner of the 3D printing/reprap revolution. With some delay, these sort of processes are now starting to take off, as a generation of young engineers liberated from the large-scale industrial ethos work to see how low the cost of general robotics can go.
So it was that before leaving the US, I visited the Berkeley campus in San Francisco and its inevitably named Fab Lab. “Fab Labs” are makerspace-like outfits attached to university engineering and science departments, often with a startup “incubator” service attached. Fab Labs are focused on fast-tracking research from faculties into developed products. The Berkeley Fab Lab is as bright and poppy as some new Brooklyn Gawker-knockoff office started nine minutes ago, but behind the ’70s retro wallpaper and foosball tables, there’s just another bunch of grey concrete rooms.
There you find DASH robotics, Nick, Andrew and Kevin, three kids — they all have PhDs, but they look 14 — all gathered round a boxy little … well, it’s a sort of small box about the size of your two cupped hands, with a half-dozen legs sticking out. This is DASH, which, when started up with a flick of the switch at the top, takes a run for the edge of the table, plunges off it, hits the floor and keeps running. DASH’s six legs don’t really walk in that cautious robotic stroll, they sort of galumph at a furious pace, paddling away, the circular motion propelling them forward at a furious rate. “It goes at six miles per hour,” says Nick. I look a little disappointed. “You have to scale it up. If it was the size of a car, it’d be breaking land speed records.”
The secret to DASH’s speed and agility is its use of animal propulsion systems, with complex cycles and rhythms, rather than linear processes. The company’s three principals met at Berkeley’s Biomimetic Millisystems Lab, where engineers have been taking apart animal movement systems for a number of years. Crowdfunded after a prototype, DASH now has an oversubscribed list for its initial product, which, when it rolls off the line, will be mailed out. Not as a parcel, but in an envelope. For DASH is made of paper.
Well, not paper exactly, a sort of coated cardboard. The body and legs of DASH come in a printed-out sheet, like the model planes of old. The legs are assembled as a simple crossover, but then generate complex motion. DASH is powered by a simple sub-arduino board, with an off-the-shelf motor attached. “Eventually we’ll have a battery that prints out, and we’ll get every element down flat,” says Andrew. The current DASH will retail for $70. “I presume you won’t tell me what the production cost is.” “Yes!” says Nick, smiling. “We won’t! But there’s a mark-up.” They’re more or less turning out the new model by hand. When a smoother production system is in place, you could probably turn them out for $5 or $10.
Were you to mistake DASH for a toy, that wouldn’t be too impressive. But its rolling, fluid, reactive style of movement makes it so much more than that. Fitted with a simple sensor, DASH could be steered by a smartphone, individually or en masse. That vastly expands the uses of such a simple robot: for large-scale painting and cleaning, for example, mass searching in natural disasters, swarm-style data collection — especially in fast-moving environments like bushfires (I suggested that one) — and many other uses. It’s a fine example of the other sort of robotics — cheap, reproducible, easily distributed, capable of being turned out, adapted, improved from a thousand makerspaces and small operations further down the track. And though even these basic creatures cannot help but generate a degree of zoomorphic feelings — they look so busy! where are they scurrying to! — they’re not going to be something you can get confused about the need to have some sort of relationship with. You’re not going to see them at Innorobo anytime soon.
But things like these, and modified forms of the FANUC “M”, and much more besides are going to be everywhere eventually, and sooner than you think.