Flying robot promises bird's-eye view for surveillance

Flying robot promises bird's-eye view for surveillance
Portland, Ore. — If you'd watched the fumbling attempts to re-create the first airplane during the celebration of the 100th anniversary of human flight last year, you might have thought little progress had been made between 1903 and 2003. But if you had watched the real birds flocking around the robotic birds being flown over the University of Delaware recently, you'd think otherwise.

The "birds" built by mechanical-engineering professor Sunil Agrawal and his team at Delaware advance the capabilities of unmanned aircraft, which until now have been fixed-wing designs. During a test flight of the first of Agrawal's prototypes to use a battery (rather than a rubber-band mechanism) to power the wings, "nearby birds came and circled around it," he said. "That flight was only 2 minutes, but the reaction of the real birds convinced us that were on the right track."

The robotic birds herald an era of natural-looking surveillance craft that blend in to their surroundings. Specifically, Agrawal is studying hovering birds and insects. "If we can provide a stable enough platform for a miniature camera, then we could fly flocklike formations of robotic birds, each sending its surveillance data back to a central point," he said.

Besides the obvious military applications, the robots could be used in search-and-rescue missions, to map hazardous areas (such as the interiors of collapsed buildings) and to provide agricultural and industrial telemetry.

Agrawal's first robotic birds resembled the rubber-band-powered balsa wood airplanes that many people built in their youth. But Agrawal's robotic birds had flapping wings and thus dispensed with the need for a propeller. Stretched over the balsa wood frame was paper that mimicked the "skin" of an airplane's wing and stood in for the feathers of an actual bird wing. The first model weighed in at more than 1.75 ounces (50 grams).

After Agrawal's prototypes sprouted battery-powered flapping wings, real birds began to take an interest in his craftsmanship. The early battery-powered robotic birds, however, were still heavy and lacked a means of steering so instead merely circled — a sufficiently authentic birdlike action to attract the real birds' attention, but less than ideal for some of the applications for which the robotic birds are envisioned.

Agrawal refined the steering mechanism and switched from the balsa and paper structure to one built of carbon-fiber composite and Mylar, thereby dropping the overall weight to just half an ounce. His goal is to craft a palm-sized version — roughly the size of a hummingbird — that could perform surveillance missions in flocks that would be indistinguishable from the real thing to all but the most technically sophisticated eye.

A crucial step is to get the robotic birds to mimic flocking behaviors. Agrawal claims that much of the groundwork for that step has been laid, since many of the necessary algorithms for coordinated effort have already been amassed for use with land-based robots. "We want to take our experience with coordinating land-bound robots and adapt it for cooperative flying machines," he said.

Thus far, the group has analyzed the mechanisms by which birds and insects attain flight with flapping wings by studying, for instance, slow-motion video of a hummingbird. Now Agrawal plans to use a wind tunnel so that data on flying attitudes, force and torque can be characterized. "With data from a wind tunnel we can refine our designs using computer models and build better robotic birds," he said.

Funding for Agrawal's research has come from the U.S. Air Force, the National Science Foundation, the National Institute of Standards and Technology and the National Institutes of Health.