Essays. FSF Columns

In a recent US military wargame, the Blue or Friendly commander was allowed to fortify his position with experimental smart mines, unma

This demonstrates that "dumb machines" needn't be very smart at all to be of real military advantage. They don't even necessarily have to be used in battle -- the psychological advantage alone is very great. The US military benefits enormously if can exchange the potential loss of mere machinery for suffering and damaged morale in the human enemy.

Among the major robotics initiatives in the US arsenal today are Navy mine-detecting robots, autonomous surveillance aircraft, autonomous surface boats, and remotely-piloted "humvee" land vehicles that can carry and use heavy weaponry. American tank commanders are especially enthused about this idea, especially for lethally dangerous positions like point-tank in assaults on fortified positions.

None of these military "robots" look at all like a human being. They don't have to look human, and in fact work much better if they don't. And they're certainly not programmed to obey Asimov's Three Laws of Robotics. If they had enough of a "positronic brain" to respect the lives of their human masters, then they'd be useless.

Recently there's been a remarkable i

These bug robots are basically driven by simple programs of "knee-jerk reflexes." Robot bugs have no centralized intelligence and no high-level programming. Instead, they have a decentralized network of simple abilities that are only loosely coordinated. These robugs have no complex internal models, and no comprehensive artificial "understanding" of their environment. They're certainly not human-looking, and they can't follow spoken orders. It's been suggested though that robot bugs might be of considerable commercial use, perhaps cleaning windows, scavenging garbage, or repeatedly vacuuming random tiny paths through the carpet until they'd cleaned the whole house.

If you owned robot bugs, you'd likely never see them. They'd come with the house, just like roaches or termites, and they'd emerge only at night. But instead of rotting your foundation and carrying disease, they'd modestly tidy up for you.

Today robot bugs are being marketed by IS Robotics of Cambridge, MA, which is selling them for research and also developing a home robotic vacuum cleaner.

A swarm of bugs is a strange and seemingly rather far-fetched version of the classic "household robot." But the bug actually seems rather more promising than the standard household robot in 1993, such as the Samsung "Scout-About." This dome-topped creation, which weighs 16 lbs and is less than a foot high, is basically a mobile home-security system. It rambles about the house on its limited battery power, sensing for body-heat, sudden motion, smoke, or the sound of breaking glass. Should anything untoward occur, Scout-About calls the police and/or sets off alarms. It costs about a thousand dollars. Sales of home-security robots have been less than stellar. It appears that most people with a need for such a device would still rather get themselves a dog.

There is an alternative to the no-brain approach in contemporary robotics. That's to use the brain of a human being, remotely piloting a robot body. The robot then becomes "the tele-operated device." Tele-operated robots face much the same series of career opportunities as their brainless cousins -- Dirty, Dull and Dangerous. In this case, though, the robot may be able to perform some of the Dull parts on its own, while the human pilot successfully avoids the Dirt and Danger. Many applications for military robotics are basically tele-operation, where a machine can maintain itself in the field but is piloted by human soldiers during important encounters. Much the same goes for undersea robotics, which, though not a thriving field, does have niches in exploration, oceanography, underwater drilling-platform repair, and underwater cable inspection. The wreck of the *Titanic* was discovered and explored through such a device.

One of the most interesting new applications of tele- operated robotics is in surgical tele-operations. Surgery is, of course, a notoriously delicate and difficult craft. It calls for the best dexterity humans can manage -- and then some. A table-mounted iron arm can be of great use in surgery, because of its swiftness and its microscopic precision. Unlike human surgeons, a robot arm can grip an instrument and hold it in place for hours, then move it again swiftly at a moment's notice without the least tremor. Robot arms today, such as the ROBODOC Surgical Assistant System, are seeing use in hip replacement surgery.

Often the tele-operated robot's grippers are tiny and at the end of a long flexible cable. The "laparoscope" is a surgical cable with a tiny light, camera and cutters at one end. It's inserted through a small hole in the patient's abdominal wall. The use of laparoscopes is becoming common, since their use much reduces the shock and trauma of major surgery.

"Laparoscopy" usually requires two human surgeons, though; one to cut, and one to guide the cable and camera. There are obvious potential problems here from missed communications or simple human exhaustion. With Britain's "Laparobot," however, a single surgeon can control the camera angle through a radio-transmitting headband. If he turns his head, the laparoscope camera pans; if he raises or lowers his head it tilts up and down, and if he leans in, then it zooms. And he still has his hands free to control the blades. The Laparobot is scheduled for commercial production in late 1993.

Tele-operation has made remarkable advances recently with the advent of fiber-optics and high-speed computer networking. However, tele-operation still has very little to do with the classic idea of a human-shaped robot that can understand and follow orders. Periodically, there are attempts to fit the human tele-operator into a human- shaped remote shell -- something with eyes and arms, something more traditionally robotlike. And yet, the market for such a machine has never really materialized. Even the military, normally not disturbed by commercial necessity, has never made this idea work (though not from lack of trying).

The sensory abilities of robots are still very primitive. Human hands have no less than twenty different kinds of nerve fiber. Eight kinds of nerve control muscles, blood vessels and sweat-glands, while the other twelve kinds sense aspects of pain, temperature, texture, muscle condition and the angles of knuckles and joints. No remote-controlled robot hand begins to match this delicate and sophisticated sensory input.

If robot hands this good existed, they would obviously do very well as medical prosthetics. It's still questionable whether there would be a real-world use and real-world market for a remotely-controlled tele-operated humanlike robot. There are many industrial uses for certain separate aspects of humanity -- our grip, our vision, our propensity for violence -- but few for a mechanical device with the actual shape and proportions of a human being.