As a chittering cloud swirled above downtown, swelling rapidly until it was larger than a city block, the former nuclear engineer hurriedly set up three video cameras, hanging a walkie-talkie on each so that he could talk to it, then aiming the lenses at just the right place in the sky.
The nightly display of group aerobatics had begun. The flock of chimney swifts grew as the light faded until there were 4,000 of them, moving back and forth, spinning one direction, then the other, billowing and undulating, pushed by unknowable instincts, by each other’s whims and by the gusting wind.
Then, as if signaled, the birds spun into a kind of feathered tornado that funneled down a dormant office building chimney, somehow without a single crash.
“There is some order to it, some disorder to it,” said the researcher, Dennis Evangelista. “And how the two relate can almost make your mind explode.”
Never miss a local story.
It was a startling and primeval sight. And it also may be the key to one of the toughest technical challenges of modern aviation: how swarms of tiny drones could be programmed to fly in flocks without destroying each other.
Role model for drones
Such swarms could be hugely valuable for civilian missions such as searching for survivors of natural disasters, or military ones such as scouting short distances or even inside buildings. The drones in such a flock would be too numerous to be piloted individually and to a great degree would have to fly themselves.
That’s why Evangelista and his cameras were sitting atop a downtown parking deck, as they had most nights since early summer.
He has been videotaping the swifts’ nearly indescribable aerial interactions for computer analysis over the winter, using a walkie-talkie as the cameras roll to add crucial audio commentary about things like the wind speed and direction at certain points in the filming.
He also periodically keys his microphone, which makes an almost bird-like trill to help later with synchronizing the cameras, something that will ensure useful three-dimensional data about the flock’s movements.
The work is part of a five-university project for the U.S. Navy. UNC-Chapel Hill’s part was awarded to Tyson Hedrick’s biology lab, where Evangelista, a former Navy nuclear engineer, is a postdoctoral researcher. The other universities are studying things like insect flight stability and actual drone construction.
Mysteries of the flock
The swifts seem designed for the study. They are balletic, unusually skilled fliers, and the flocks form at the same time and place every night during the summer and early fall before – in the case of those here – flying to Brazil for the winter.
The scientists hope they can decode the movements of the flock. If they can begin understanding how swifts do things like learn from the motions of birds 50 spaces away in the flock to make a good approach to the chimney, that kind of crowdsourced approach to close-formation navigation may be transferable to drones.
Eventually it got so dark that the naked eye couldn’t pick out individual swifts, just the swirling motion of the cloud and a dark smudge above the chimney that looked like smoke streaming backwards. When it was too dark to see the last few birds, Evangelista began packing away his cameras and tripods.
The precise reason that chimney swifts swarm like this may forever be a mystery, one that Evangelista – as much as he likes to find answers – appreciates.
The species existed, of course, before there were chimneys, and so good roosts – suitable hollow trees – may once have been so scarce they had to share, Evangelista said. Another reason could be that banding together spreads the risk for any one bird of being picked off by a predator such as, say, a hawk.
“Biology is fun,” he said, “because you get some questions that never end.”