Analog Science Fiction and Fact 11/01/10

by ASF

  “It’s a very simple concept,” one of the inventors, Mitch Tyler, told On Wisconsin, the alumni magazine of the University of Wisconsin, Madison. “It’s like having someone place a finger on top of your head to indicate you’re upright. If you tip your head, you feel the finger slide off to one side, and you naturally move your head back to compensate. You’re just correcting for a deviation in your position relative to a marker.”

  It worked. Amazingly well, and amazingly quickly—within minutes, in fact. “Once the concept is in place that the stimulation on the tongue means something about your orientation in space . . . it is very intuitive,” Tyler said. “It very quickly goes from being a conscious process to being subconscious.”

  The device, now sold as BrainPort, is an early predecessor to Kip McCorbin’s Sense.

  In the story, Kip uses a tattooed implant to pilot a swarm of cyborg insects equipped with microsensors that give him an ability to see and hear at a distance. Lie-detectorlike sensors also let him read emotions to a range of about 500 meters, helping him detect danger. (The range is limited by the ability of the insects to carry batteries powerful enough to relay their transmissions.)

  There are a number of technologies involved, but one of the most difficult-sounding ones—interpreting feedback from the swarm—might be among the easiest. That’s because, as the success of BrainPort indicates, the brain is remarkably “plastic”—meaning that even in adults, it can be molded to reinterpret the stimuli it receives.

  Nor are we limited to using the tongue. Any convenient patch of skin should do. Someday, similar technologies might allow rehabilitation clinics to link cameras or microphones to an output grid that gives touch-based sight to the blind, or hearing to the deaf. These wouldn’t truly be sight or hearing, but to those trained to recognize them, they might seem very much like the real thing, except, perhaps, for differences in resolution due to the comparative “graininess” of touch receptors compared to the retina or inner ear.

  Kip uses the term “integration” to describe the process of learning to interpret such signals. In his case, with 300 insects providing multiple types of feedback, it would be more difficult than simply trying to position your head so that an electrical buzz stays centered on your tongue. But the concept is the same. “If you’ve got the ability to integrate,” Kip says, “the data cease to be data . . . [Y]ou wind up with things you simply know, on par with it’s raining or I’m on a tropical beach.”

  Controlling the insects also sounds fanciful, but on a limited scale, it’s possible today. In 2009, a team lead by Hirotaka Sato of the Department of Electrical Engineering and Computer Science at the University of California, Berkeley, reported that they had developed an implant that allowed them to remotely fly a beetle, using a laptop computer as controller. (Video of the flying insect is available online.)

  The technology, called HI-MEMS (for hybrid insect micro-electro mechanical systems), uses electrodes implanted into the beetle’s nervous system and muscles during its pupal stage. As the beetle metamorphoses into an adult, its brain and muscles integrate with the electrodes, which can then be controlled with a chip and microbattery mounted onto the insect’s back. WiFi-style signals to the control chip can then be used to hijack the beetle’s control of its wings, allowing the controller to direct its movement.

  It sounds a bit spooky. But a press release from the Defense Advanced Research Projects Agency, which has gone public with plans to use the beetle as a platform for sensors such as microphones or gas sniffers, argues that it’s not all that strange. Rather, the agency suggests, it’s a lot like what we’ve long done with other animals by lower-tech methods like yokes for oxen or bits and reins for horses.

  Nor are implanted chips and WiFi the only ways to take control of an animal’s nervous system. At the 2010 annual meeting of the American Physical Society, March 15-19 in Portland, Oregon, Andrew Leifer, a biophysics graduate student at Harvard University, showed videos of how he had been able to seize control of a tiny worm, called Caenorhabditis elegans, using green laser light.

  C. elegans was chosen because it has only 302 neurons—and because it’s transparent. The latter is important because the key to the research was a gene modification to the worm that allows researchers to activate its neurons by exposing them to green light. Shining light on the entire worm activates everything at once. But training a tiny laser spotlight on a single neuron activates only that specific one. At this point, it’s simply a research project, designed to figure out how the one-millimeter-long worm’s nervous system works. “This is going to be a really powerful tool that will allow us to probe neural circuits with unprecedented specificity . . .” Leifer said at a press conference. “We now have the power to systematically decompose a [neural] circuit and tweak each component to see what it does.”

  But he can also seize control of the worm’s movements, making it reverse course, turn right, or turn left. “You could play the worm like a piano,” one reporter noted.

  The hardest part, Leifer says, was focusing a neuron-sized spotlight on a tiny, wriggling worm. But eventually, he got an assemblage of 700,000 independently controlled mirrors to direct his beam whereever he wanted in a mere 20 milliseconds. “I spent six months writing code,” he adds.

  Will this technology supersede HI-MEMS for controlling cyborged insects? Who knows? Leifer’s experiments relied on the fact that C. elegans is transparent. But that’s not necessary for laser control, another brain researcher at the meeting noted. Instead of activating gene-modified neurons by shining green light on them, he said, all that would be needed would be to insert electrodes into the insect’s brain, with each one attached to a tiny photocell. The desired neurons could then be activated with beams of light. Such a method would have the advantage of using signals that don’t produce radio leakage detectable by enemy monitoring stations, but it has the enormous drawback of requiring both line-of-sight and a very fast-tracking laser. Most likely, WiFi would be the preferred option.

  Bug Power

  In today’s cyborg experiments, insects are controlled one at a time with the cursor keys on a laptop computer. But Kip controls hundreds at once, via a tattoolike implant on his body. Again, several technologies are involved.

  To begin with, the implant draws on his metabolic processes to recharge the batteries of his insects’ micro-controllers when they run low—something that will probably happen fairly often due to limits on the size of batteries an insect can carry.

  Not that the insects are likely to carry conventional batteries. More likely, they’ll have ultra-capacitors, which store charge, rather than producing it chemically, as batteries do. Large capacitors, already in use in China, can power a city bus, reputedly for fifty kilometers between rechargings, says Saikat Talapatra, a condensed matter physicist at Southern Illinois University. But even better ones are in the works. Talapatra’s group is working with single-atom-thick sheets of a super-strong carbon material called graphene, capable of holding massive quantities of charge per gram. Such capacitors also have the potential of being recharged very quickly—just as Kip does in the story by hovering his bugs close to power stations on his tattoo.

  The ultimate source of energy, however, is Kip’s own biochemistry. And that, rather startlingly, involves a technology that’s already available, at least in the design stage.

  In 2008, the Greener Gadgets Design Competition announced the development of a tattooed cell phone, “powered by pizza.” Technically speaking, the phone is powered by cheese or pepperoni. Rather, it runs on glucose, via a coin-sized fuel cell that taps into the user’s bloodstream. As blood flows through the fuel cell, it extracts glucose and oxygen to generate electricity.

  The cell phone itself is composed of two pieces. One is a thin, flexible sheet of silicone inserted beneath the skin. That’s the actual electronics, powered by the blood-fed fuel cell. The other layer is the tattoo itself, which is invisible until it’s needed, at which point signals from the silicone im
plant light up a keypad and view screen. (Sound presumably arises directly from the implant.)

  Most of us might not want to be that closely connected to our phones—but if we can do this, we can definitely find ways to harness Kip’s blood to recharge his swarm.

  Swarm Intelligence

  Controlling the swarm is a bit more difficult. The basic principle is the same as the one used to “integrate” incoming data, but in reverse. Here, rather than commandeering sensory nerves to receive incoming data, it’s necessary to use motor nerves to relay signals to the insects.

  The basic technology is already under development for restoration of fine motor control to prosthetic hands. Numerous systems are possible, but one of the most interesting would use artificial neuromuscular connections at the ends of the severed nerves, allowing them to power servomotors in the prosthetic.

  Kip is going to need something similar, although we won’t want to disconnect the nerves from their normal uses, so we’ll have to tap into their signals without blocking them. In theory, the best muscles to use might be those controlling his hands. That would certainly give him the greatest possible range of control, but it would also mean that waving, pointing, or scratching his nose might send his insects zinging off in random directions. Not to mention that if he walks around moving his hands all the time, he’s going to stand out in a crowd.

  A better approach is to use small muscles in the chest—a technology already in use for prosthetic hands. When the user thinks, “Open the hand,” or “Flex the index finger,” the thought actually activates these chest muscles. But electrodes monitoring them pick up the signals and relay them to the artificial hand. It’s another example of brain plasticity in action . . . and a rapidly maturing technology, already good enough, advocates say, to allow users to play the piano.

  Controlling one insect this way would be easy. The Berkeley lab group that first cyborged a beetle appears to have used only six commands: take off, land, turn right, turn left, climb, descend. Multiply that by 300 insects, however, and Kip has to be able to execute up to 1,800 distinct commands . . . potentially all at once.

  Or does he?

  If each insect is independently controlled, he has a problem. And he’s definitely going to need the ability to control (and monitor) them one at a time when he wants to. But for general-purpose movement, he can get a big assist from a programming concept known as swarm intelligence.

  Let’s start with a very simple example: pedestrians on a sidewalk. At the 2010 meeting of the American Association for the Advancement of Science, February 19-22 in San Diego, California, Mehdi Moussaid, a mathematician at the Swiss Federal Institute of Technology, described experiments in which he filmed pedestrians on a sidewalk and catalogued their behavior. In zero congestion, he found, groups tended to walk line abreast. In light crowds, they tended to form V-shapes, with the people on each end moving slightly forward. In more crowded conditions, the V inverted, with the center person taking the lead and others following off to each side, like a flock of geese. And in extreme conditions, groups moved single file.

  Does anyone tell them to do this? No. It’s the result of two simple motivations on the part of each person acting separately: (1) they want to get from point A to point B; and (2) they want to be able to talk to each other as much as possible, en route. The shifting patterns come about simply as each reacts to what’s going on in the immediate vicinity. Line abreast is the most efficient pattern for social interaction, but it’s the least efficient for moving in a crowd. A V-pattern still promotes social interaction, but when the people on the ends start encountering too much jostling, they drop back, and the V spontaneously inverts. Single file comes about when congestion is so extreme that conversation is nearly impossible.

  Even more interesting is what happens to solo pedestrians moving in heavy congestion. When they meet head-on they have a tendency to side-step preferentially to one side—right in the U.S.; left in some other countries—presumably controlled by which side of the road cars drive on. The result: suddenly, a crowd can spontaneously reorganize from chaos into lanes, just like highways.

  People aren’t the only ones who do this; streams of ants going in opposite directions will also form lanes (as will ball bearings rolled at each other from opposite directions), although in this case there won’t be the culturally driven “proper side of the road” bias. What’s interesting is that it occurs with no overriding control. Rather, each person is pursuing the path of least resistance, with a slight bias toward sidestepping to the culturally preferred side.

  Kip’s swarm isn’t composed of pedestrians on a sidewalk, but some of the same underlying principles apply. “You have a high number of agents, and somehow the whole group manages to organize,” Moussaid says, speaking specifically of pedestrians, although the remark applies more broadly. “Individuals having [only] local perception of their environment manage to do something together.”

  Another term for this is emergent behavior.

  At the Monterey Bay Aquarium, in Monterey, California, it’s possible to watch sardines swimming in a giant tank. One moment, they’re all swimming in one direction, a united, purposeful-seeming entity. Then suddenly, the entire school changes course. But try to find the fish that initiates the change! Rather, it seems as though the entire school has made a collective decision: we’re tired of going this way: let’s try something new.

  The fish aren’t acting in concert; nor are they following a leader. Rather, each is reacting to its neighbors according to fairly simple rules, so quickly that the “decision” to switch courses seems to have been made collectively. The flocking behavior of birds is similar. And it’s easy to mimic. In lecture notes for an introductory robotics course, Maja J. Mataric, director of the Center for Robotics and Embedded Systems at the University of Southern California, describes a simple procedure for creating robots that flock like birds (or school like fish). All that’s needed is for each robot to follow three simple rules:

  • Don’t collide with another robot.

  • Don’t get too far away from other robots (with “too far” being defined more precisely in the actual programming).

  • Keep moving.

  That’s it. This won’t control where the flock goes, but it will cause the robots to group and move as a unit.

  Another simple type of emergent behavior is what Mataric calls “wall following.” This involves a single robot, interacting with an environment about which it has very little information. Here, the instructions are:

  • Move at random until you encounter a solid object.

  • Don’t get too close to it.

  • Don’t get too far away, either.

  • Keep moving (but don’t simply oscillate back and forth).

  The result: a robot that prowls walls, either in two dimensions (if it’s a wheeled robot rolling across a floor) or three (if it’s a cyborged insect).

  Obviously, Kip doesn’t want his swarm to behave like a school of fish. And simply following walls is a bit boring. But the same basic approach can give us more complex behaviors.

  If we wanted, we could combine the wall-following and flocking algorithms to produce a flock that follows walls. With the right constraints on each robot’s optimum distance from the walls and each other, we could get them to fan out across the entire face of a building, letting their onboard cameras and other sensors peer through all the windows. Or we could program them to recognize open windows or other gaps, so that when one finds a way in, it could lead the others inside, too.

  In other words, given a few simple rules, the members of the swarm can be programmed to react to each other and their environment, forming complex, seemingly coordinated behaviors, without the need for Kip to control them one by one.

  All of this makes Kip’s job a lot easier. Rather than trying to fly all 300 insects individually, he simply divides them into task forces and gives each group a simple instruction, such as “fan out,” “establish a forward perimeter,”
“hide,” “monitor object X,” “find a way into that building,” etc. These assignments activate various algorithms that would let the swarm, or a task force, carry out much of the job on its own, while Kip looks on, occasionally taking control of individual insects if he needs to do something specific.

  That sounds complex, but remember what we said earlier about brain plasticity. When one of us (Richard) was in eighth grade, his parents made him take a typing course. At the start, everything was one letter at a time as he searched the keyboard for the right keys. S o m e t h i n g l i k e w r i t i n g l i k e t h i s. Now, typing is nearly automatic. He thinks, and the words appear on the screen while he performs very complex movements at up to 100 words per minute. If he tries to think too much about what he’s doing, in fact, he freezes up. It’s like walking. He just does it.

  Any time we do a skill like that—whether it’s typing, shooting hoops, or playing fast-acting computer games, we’re using the same basic skills Kip uses to pilot his swarm. And just like Kip and his fellow CI-MEMS operators, some of us are good at it, some aren’t, and some get addicted.

  Copyright © 2010 Richard A. Lovett & Mark Niemann-Ross

  Short Stories

  Short Stories

  The Zoo Team

  Allen M. Steele

  We were somewhere over Australia, about a quarter of the way to Mars, when Miguel flipped out. Ron and I had a lot to do with his breakdown, and when it was all over we were quite proud of ourselves. A good, full-blown mental collapse takes time and effort, of course, and we’d spent the last...

  Contamination

  Jay Werkheiser

  Ari allowed his skimmer to brush the outer edge of Nouvelle Terre’s atmos-phere. He tried to imagine air jostling the light nanofiber support frame, whistling through the skimmer’s magsails....

  The Deadliest Moop