The biggest challenge facing MAV designers is the same problem encountered by early pioneers of manned flight: finding a sufficiently lightweight powerplant. Piezoelectric motors are efficient, but still require electricity. Wings covered with photoelectric materials work to a degree, but ultimately reach the point at which they have too little light-capture area to produce enough power to flap wings or run electronics. Batteries can provide power for short flights, but are too heavy for missions extending beyond a few minutes.

The consensus among experts interviewed by Popular Mechanics is that MAVs will need to produce their power on board. There are three possible technologies. The most powerful is the 13mm Microjet demonstrated by the British Defence Evaluation and Research Agency (DERA) at this year's Farnborough International 2000 air show. "By mixing hydrogen peroxide with kerosene or a similar fuel, we've achieved flight duration times of up to 1 hour," says a DERA spokesman. "Starting and stopping the engine is very simple and is achieved by a simple on/off value, making it reliable and simple to operate in the field."

MIT's Epstein is trying a somewhat different approach. His lab has received about $5 million from the Army Research Office to develop a microturbojet that, ultimately, could be mass-produced using the same tools and techniques used to make computer chips.

Like a conventional jet engine, the MIT miniturbine would have a combustion chamber, turbine wheel and compressor wheel. Fuel burning in the combustion chamber would send exhaust gases through the blades of the turbine wheel, causing it to rotate and, in turn, drive the

compressor wheel via a central shaft. Small as they are, the DERA and MIT jet engines will be like Saturn V rockets compared to the motors being developed at Georgia Institute of Technology in Atlanta. "We are now being driven by fundamental, technological and economical considerations to explore and evaluate systems that are smaller and smaller," says Uzi Landman, director of Georgia Tech's Center for Computational Materials Science. To study tiny nanojets, Landman and collaborator Michael Moseler are using molecular dynamics simulations to observe the furthest frontiers of fuel economy-combustion involving as few as 200,000 propane molecules.

What might seem to be the most formidable challenge in developing MAVs-getting tiny robot fighters to find their targets-is actually less of a problem than one might think. "Microelectronics technology is the driving force behind shrinking systems. On-board computational capabilities per unit volume will continue to increase," says Epstein. He points to shrinking avionics, which include GPS receivers weighing as little as 6 grams, roughly the weight of 12 aspirins. Getting instructions into MAVs is also fairly straightforward. Infrared (IR) ports, like those on personal digital assistants, will allow MAVs to be programmed in the field. And, once in the battlefield, IR ports can be used to send coordinating instructions within a swarm.

A major breakthrough that will make these systems even smaller was recently reported by chemists at the University of California at Los Angeles (UCLA). They have coaxed ringlike groupings of rotaxane molecules to exhibit the on/off behavior of transistors. Industry experts who have examined this process say it has the potential to put the computing power of 10 Pentium processors in one-hundredth the space of one of these tiny chips. And, because rotaxane molecule transistors could be switched on and off using light, there would be no bulky wire interconnections needed. MAVs will someday carry the computing power of an F-22.

The merger of Georgia Tech's nanojets with UCLA's molecular transistors might bring about a newer, even smaller and smarter class of robot warriors: nano-MAVs so small they could hide on the wings of real flies.