In science fiction, we take it for granted that alternate realities can be touched and that robotic sensors will work just like our own. When Luke Skywalker loses an arm, he tacks on another that responds so much like the original that his Jedi sword never falters. Meanwhile, on the holodeck, Star Trek regulars pick up and set down the clues left by a computer-generated Dr. Moriarity.
If in our own space-time continuum things lag a bit behind, it isn't the fault of specialists in the field of haptics, who work with touch-related human-computer interfaces.
The study of haptics, from a Greek term relating to "touch," embraces not only computer input devices that rely on human hands, feet, or mouths, but also output devices that allow the computer to provide tactile sensory data to a human on the other side of the device. This kind of force-feedback device allows you not only to use your joystick to push your way through the virtual storm you encounter in your gaming, but also to feel in the joystick the reverberation of the thunder or the sudden letup of force as you take shelter from the wind.
Haptics is a burgeoning area of academe, sponsoring learned conferences this year from Chicago to Munich, to Pisa. But how far has it brought us along that imaginary road on which we can lean down and lift a computer-spawned sixpence from the wet pavement and run a thumb along the smooth, cold edge?
A Few Obstacles
The study of haptics is not all gaming and globe-hopping. There are some unsettling challenges.
To begin with, the amount of information we take in from touch is staggering. Run your finger across your cat's coat, and you pick up a wealth of data from the point of contact. The fur is soft or harsh, wet or dry, long or short. But these sensations are really compiled evaluations of the information we receive. Our fingerpads are so sensitive to texture that we can discern a variation on a smooth surface that is only 100nm high, according to Karen Birchard, writing in The Lancet.
Now, you might think the recognition and evaluation of this kind of minute variation is hard to reproduce, and you would be right. However, Tekscan markets sensors that depend on layered silver electrode arrays, and, using this technology, sensors have been made recently with spatial resolution as fine as 0.0229 mm2.
And while that sounds as if we're closing in on the sensitivity issue, fingerpads aren't the only thing communicating tactile information to our brains as we ruffle that cat's fur. We depend on things like resistance to know how hard we can run our fingers down the back of Thomasina's neck. Such information comes not just from our fingertips, but from joints and tendons as well. Without that information to alert us to how hard we should press, a protest from kitty would probably be promptand, to the human hand, possibly painful.
Creating robotic hands that respond as human hands do depends upon more than sensors. Moore's law, which says the computing power per square inch of printed circuitry doubles every 18 months, does not apply to the motors that move mechanical hands and robotic arms. Furthermore, resistance in haptic devices is provided by magnets, which cannot be made smaller but stronger.
These issues constitute something of a roadblock in robotics, making it tricky to provide a cheap and reliable android who can wash the dog for us. The only rational conclusion is that if we want Rover to be clean, we'll be running the bathwater and floating his rubber duck ourselves for some time to come.
Another mechanical demon is that when you push your computing devices and they can push back, someone could get hurt. While this isn't much of a worry with a force-feedback touchpad (most of us figure we could take one on bare-handed, should it come to a showdown), it is a consideration in therapeutic applications in which a malfunction could mean that the wrong sort of torque at the wrong time damages otherwise healing tissues, and it is a concern in industrial applications in which some haptic interfaces may need plenty of punch at their disposal.
One technology that promises a solution is magnetorheological (MR) fluids, under study at Georgia Tech. "If MR fluids are the sole mode of actuation for a haptic interface, then risk of human injury is significantly reduced," says Matt Reed, a control systems engineer at Northrop Grumman Space Technology.
A system actuated entirely by MR fluid brakes or dampers is energetically passivethat is, it can only resist or redirect motion. As a result, all motive energy must be applied by the human operator, a factor that increases safety and guarantees system stability. "In the case of an active system, the device could be augmented with MR fluid actuators in order to increase safety," says Reed, cautioning that this adds cost and complexity. "The additional damping and reduced reliance on the active actuators to generate force could reduce the potential risk of injury."
These studies impact how quickly we have commercial applications. Until researchers and legal departments are both sure they have adequate safeguards in place to keep your 200-lb. pup from being harmed, you won't be getting that XPE-404-9 Household Bot that's sturdy enough to lift your Neapolitan Mastiff and toss him into the bath suds.