Microfly Generates Research Buzz

Multidisciplinary approach drafts unlikely candidate for missions.

A research pipeline between biologists and engineers has led to a new class of microrobotics, spawning a paperclip-sized mechanical flying insect that will weigh one-tenth of a gram and will measure 1 inch from wing tip to wing tip. The result will be applied in search and rescue missions, mine detection and even planetary exploration.

Drawing from experience in the emerging field of biomimetics, researchers are creating these tiny flying robots by looking at nature’s designs such as the fruit fly. The five-year, $2.5 million project involves studying how the insect flies, navigates and searches. This joint effort of the Defense Advanced Research Projects Agency (DARPA) and the Office of Naval Research (ONR), both in Arlington, Virginia, is providing advances in communications, control tools and miniaturized autonomous systems.

The two sponsors of the project have been looking at a number of ways to implement small, practical mobile sensor systems. The robotic menagerie also includes a moth, lobster, lizard, ant and octopus, which are being researched by members of private institutions and academia. Scientists plan to mimic these creatures to develop microversions of propulsion, power and electronics.

According to Michael H. Dickinson, professor, department of integrative biology, University of California–Berkeley, it was the fly that most intrigued ONR scientists. At the same time, DARPA officials approached him about submitting a proposal to study insects to gather ideas for control systems. It was clear that both agencies’ efforts could complement one another.

Dickinson notes that one of the biggest challenges of building a micromechanical flying insect (MFI) is creating enough force to keep the device in the air. “For a long time, insect flight aerodynamics were very mysterious, and traditional analysis led to the conclusion that insects can’t fly, which is obviously not the case,” he explains. “If you apply standard steady-state aerodynamic theory that works perfectly well for 747s and F-15s to an insect wing, the theory shows that the wing can’t generate enough force to keep the animal in the air.”

In the past few years, this mystery has been solved. Scientists now have a better understanding of how insects generate enough force to get themselves in the air. This information is a prerequisite to building a mechanical device in the same size range.

One of the ways Dickinson and his colleagues study how a fly can stay aloft is by tethering real flies in virtual reality flight simulator chambers and examining wing kinematics. The wing motion is then mapped and programmed into Robofly, a larger model of the MFI.

Robofly is immersed in two tons of mineral oil so researchers can visualize air flows and measure forces on the wings. Because surface friction is relative to size, when the movements of a live fly are mimicked by the much larger Robofly, a medium more viscous than air is needed to accurately reflect what a tiny insect experiences while moving through air.

“We use Robofly to study the unsteady aerodynamics of flapping flight—the forces that [are] generated because the wing is flapping back and forth and rotating,” Dickinson says. “The fly’s wing is doing a lot of exotic, sophisticated movements unlike an airplane wing that moves unidirectionally.” The larger mechanical device tells the engineers how a wing must move to generate enough force and how a wing’s movement must change to steer the insect. The payoff is applied to the microversion of Robofly.

Once insect wing movement data is pinpointed, the most important aerodynamic piece of the puzzle becomes creating a structure to get the robotic insect wings to move in the same way, Dickinson says. Researchers are pursuing a modified wing hinge to accomplish this goal.

The wing hinge hooks up the analog of muscles that actuate the wing and allows the wings to flap back and forth and rotate. Because the wing is small, scientists cannot use the standard types of joints that are possible at a larger scale.

Instead of a traditional hinge, flexure joints—tiny flexible strips that connect one rigid section of the insect’s skeleton to another—are used. Some parts of the skeleton are hard while others are flexible so that the wings and legs can bend. “It’s like creating the world’s most complicated origami,” he says. “The end result is that when muscles act on the structure, instead of bending it or deforming it, the wing accomplishes a marvelous motion.”

Another key technology hurdle is the development of an actuator. According to Timothy D. Sands, a professor of materials science and engineering at the University of California–Berkeley, the fly’s electrical power—currently from solar cells—must be converted to mechanical energy to create the motion of the wing.

Sands and his research team initially tried to buy commercial off-the-shelf actuators but found that the size they needed was unavailable. Additionally, the material they want to use—single crystal lead zinc niobate-lead titanate—was discovered only a few years ago and is new to the scientific community. Its piezoelectric properties are dramatically enhanced in this single crystal form, which is why the team chose the material. “We had to make actuators smaller than anyone will sell to us, and we had to use new and exotic materials,” Sands remarks.

Scientists are experimenting with thin slices of the substance to create actuators, Sands explains. To accomplish this, the crystal is bonded to steel, and voltage is placed across the crystal. The result is an actuator that is one-tenth of a millimeter thick, one-half of a centimeter long, and 2 millimeters wide.

Four actuators will be attached to the MFI’s thorax. Two actuators—working together to both cause the wing to flap and change the wing angle—will drive each wing.

Sensors have been another obstacle for researchers. Ronald S. Fearing, a professor at Berkeley’s electrical engineering and computer science department, shares that the MFI team is considering complementary metal oxide semiconductor image sensors, gyroscopes and biologically inspired visual sensors.

Additionally, Fearing plans to make use of smart dust—the product of another DARPA-sponsored project. The goal of this project is to integrate a complete sensor/communication system into a cubic millimeter package. This technology also could be applied in defense-related sensor networks or a virtual keyboard where dust motes are glued on each fingernail. The progress with smart dust will match well with that of the micromechanical flying insect, which is scheduled to fly in a laboratory setting by 2003, Fearing predicts.

Despite researchers’ anticipation of how sensors will evolve, Dickinson points out that current progress is slow. “It is generally acknowledged in the field of robotics that one of the holes in the discipline is the unavailability of sensors,” he says. To address this issue, the fly could be equipped with simplified detection capabilities instead of sensors.

“The fruit flies we work with have the equivalent of 20/4000 vision, which is legally blind for humans,” Dickinson says. “Nevertheless, they’re able to find really tiny objects with remarkable efficiency.

“Many people talk about surveillance. A mechanical insect is not going to carry a high-resolution imager. It’s not going to take picture-quality images of anything very easily. It’s much more feasible for the device to detect something like CO2 or a signature of a mine and to program the fly with simple eyes to follow an iterative search paradigm that would allow it to localize particular targets,” he explains.

This search function could be useful in finding humans trapped in rubble after an earthquake or for detecting adversaries in a building at night. It is the type of mechanism that real insects use, Dickinson explains. The animal does not require a huge computer, and yet it is still robust—apt at finding objects that are warm or are emitting an odor. “These are the types of search algorithms that you’d want to run on the MFI,” he adds.

The types of sensors and search mechanisms the MFI developers choose also will affect the mechanical insect’s control capabilities. Researchers envision enabling visual sensors that will be used by the device to detect the approach of objects and avoid collisions.

“We’re trying to figure out what makes them so robust,” Dickinson says. “What is it about their control systems that makes them fly stably? Why don’t they crash?”

Tests are being run to help answer these questions. For example, real flies are placed in what Dickinson calls arenas that allow the animals to be studied in three dimensions. Obstacles are placed in this environment, and the insects are starved briefly so they are motivated to fly around and search. The goal is to determine how changing visual environments influence the way the insect navigates.

“We also have flight simulators where we tether the insect in one spot, and we place a virtual world around it,” Dickinson relates. By changing its wing motion, the animal can control a panoramic visual display that recreates the visual pattern it would see if it flew through the air. “It’s like the fly is playing a video game. We manipulate the video game to see how the fly is processing information that enables it to fly stably in the world while exploring it.”

The military, space agencies and civilians are all in line as beneficiaries of the MFI end product, Fearing says. The mechanical insect will be constructed of steel, Mylar and polymers. Twenty MFIs will weigh the equivalent of one penny. Researchers can let their minds wander for potential uses of such a small device, he observes.

Search and rescue is one application. “MFIs could be great tracking devices. We all know this because everyone has been plagued by flies and mosquitoes,” Dickinson quips. “They’re tiny but they have a very simple and efficient search strategy. They can find warm, smelly CO2 emitters.”

An MFI with similar capabilities could be used to locate individuals who are incapacitated in an earthquake or other disaster. Dickinson anticipates that in the future every search and rescue team would have a jar of MFIs that would be released when needed. The flies could travel easily into places that rescue dogs could not.

The National Aeronautics and Space Administration also has expressed interest in using these small autonomous robots to explore planets. The flying device would be economical where the cost of transport is key.

“There is a danger in any kind of remote exploration project if you put all of your eggs in one basket such as a complicated land rover,” Dickinson contends. “If that device breaks, then you’re out of luck.” The notion of a swarm of inexpensive robots that take off in all directions to collectively gather data might be a better approach, he says. “If a few of the insects are lost—even half of them—you wouldn’t see a disaster like we have seen in recent planetary exploration. It wouldn’t end the mission.”