The researchers also demonstrated that their robot could gather materials and solder them into microcircuits, as well as use an onboard atomic force microscope probe to feel its way along a patterned surface, locating itself with an accuracy of two nanometers, which is less than the width of a DNA molecule. The probe could also be used to measure a cell's electronic or mechanical properties, and could write with nanoscale precision using a technique called dip-pen lithography.

While robot teamwork applications could be years away, the components invented for the robots might appear in products sooner than that. Already, says Seyfried, industry has expressed interest in the "syringe chip," which might find applications in labs that need inexpensive but highly accurate tools for single-cell injections. Also, new micromechanical actuators might be used in toys and consumer electronics.

But going from a couple of prototypes to large teams of robots will mean overcoming a big hurdle: energy. The Micron robots, and an earlier minirobot developed at MIT, the Nanowalker, receive power through the floor that they operate on. But including many functions in a small package can draw a lot of power -- so much that only one Micron robot at a time could operate on the power floor.

The little packages can also generate a lot of heat. The Nanowalker robot, which requires considerably more power than the Micron robot, works in a helium-filled freezer to stay cool. The freezer can be -70 degrees Celsius -- far too cold for working with, for example, living cells. Sylvain Martel, who developed the Nanowalker, is now working at Ecole Polytechnique de Montreal on the next generation of the robot, which should be both much smaller and require less energy.

Energy concerns may dissipate as integrated electronics and other innovations make the robots more efficient. Even if the energy problems are solved, though, some still question whether tiny robots are the best solution for high-precision and rapid assembly. Ralph Hollis, who develops desktop-sized robots for microassembly at Carnegie-Mellon, says the advantage of having high numbers of Micron-like robots building devices simultaneously only happens when you need something like one million devices made per day. For volumes under 100,000 or so, an assembly line approach works fine, he says.

Those who support minirobots for manufacturing, however, say their main advantage may be cost. "As you're getting smaller, the cost per robot should go down," Fearing says. "You can start to think about doing things with a hundred robots working in parallel, where at the macro scale it would be too expensive."