Vanderbilt engineers build robotic bugs that can go the distance

By David F. Salisbury
December 1, 2000

Mechanical engineers at Vanderbilt University have designed and constructed a small robotic "bug" that can scuttle more than half a mile on a single battery charge.

The leader of the project, Michael Goldfarb , reported on the development of what may be the smallest-sized robot capable of traveling significant distances at a special workshop on mobile micro-robots on Friday, April 28, at the IEEE Conference on Robotics and Automation in San Francisco.

"We set out to determine just how small we can make viable robotic technology, and we’ve done that," says Goldfarb.

Although they started out with even tinier versions, the smallest practical robot that the researchers could produce turned out to be about the size of one of those giant rhinoceros beetles found in thetropics. It is about three inches long and weighs about two ounces.

The two-and-a-half year program to develop the robotic bugs is part of a larger program funded by the Defense Advanced Research Projects Agency (DARPA) that is exploring the use of mobile micro-robots for military reconnaissance and intelligence gathering. The basic idea is that soldiers could carry large numbers of these lightweight, mechanical scouts and use them to investigate the terrain ahead, detecting enemy troops, minefields and other hazards.

To achieve a minimum size, Goldfarb and Ephrahim Garcia took a different approach. They decided to combine batteries with an unusual material called piezoelectric ceramic (PZT) that physically expands when an electrical voltage is applied to it.

For locomotion at this scale PZT had a number of potential advantages, the researchers figured. "It can be made in one piece," says Goldfarb. "There are no significant lower limits to the size of the actuators that you can make. They are also very energetically conservative: They don’t throw away a lot of energy, so most of the electrical energy comes out as mechanical energy." In this respect PZT is 90 percent efficient, compared to about 60 percent for DC electric motors.

"Two-and-a-half years ago, when we began this project, we did not know whether this alternative design paradigm would work," he says. In fact, Goldfarb and his research team, which consists of research engineers Mike Gogola and Dan Monopoli plus four graduate and undergraduate students, went through more than 10 prototypes before achieving an optimal design.

One of their major challenges was creating a mechanism that would travel at a reasonable speed. "The first few designs that we came up with had maximum speeds of about a millimeter per second, about one three-hundredths of what we can do now," Goldfarb says.

Their final design can cover about a foot per second and carry a one-ounce payload. That allows it to tote a battery that can keep it running for 45 minutes, long enough to cover about a half mile. The lightest chip video cameras available commercially weigh about half an ounce, so the insectile robot is perfectly capable of carrying one, although the researchers haven’t done so.

With a second DARPA grant, the Vanderbilt researchers have been attempting to apply this same approach to create flying robotic bugs, micro-ornithopters that fly by flapping their wings. But here the engineers may have run into a fundamental obstacle. PZT ceramics are highly efficient but they are also relatively heavy. The critical factor in mechanical flyers is not energy efficiency, but energy density the amount of energy that a device can develop per ounce.

Goldfarb’s group has designed and tested a number of different micro-ornithopters. But so far none of them has been able to lift its own weight. For the last few months, the researchers have been doing experiments to determine exactly how much power PZT actuators can deliver and to see if there are any tricks they can use to make them deliver more. So far, none of the tricks they’ve tried have worked.

While the Vanderbilt group has concentrated on the issue of locomotion, other research groups are looking at controllers that allow them to navigate and give them some degree of artificial intelligence. Still others are exploring different kinds of payloads that they could carry such as motion sensors, video cameras, microphones and other sensor-packs to detect land mines, toxic gases and biological weapons.

Proponents point out that robotic creepy-crawlers of this sort might also find a number of non-military applications in law enforcement, security, and inspection of pipes, ducts and other inaccessible or hazardous areas.

Vanderbilt Center for Intelligent Mechatronics:
http://fourier.vuse.vanderbilt.edu/cim/cim.htm

DARPA’s Microsystems Technology Office Distributed Robotics:
http://www.darpa.mil/MTO/DRobotics/index.html

Prof. Michael Goldfarb’s home page:
http://www.vuse.vanderbilt.edu/~goldfarb/persinfo.htm