To design a better drone, scientists could learn a thing or two from birds’ ability to maneuver through narrow spaces.
Budgerigars can fly between gaps almost as narrow as their outstretched wingspan rather than taking evasive measures such as tucking in their wings.
[related]
Previous research has shown that humans unnecessarily turn their shoulders to pass through doorways narrower than 130 percent of their body width. Birds are far more precise.
“We were quite surprised by the birds’ accuracy—they can judge their wingspan within 106 percent of their width when it comes to flying through gaps,” says Ingo Schiffner, researcher at the University of Queensland Brain Institute.
“When you think about the cluttered environments they fly through, such as forests, they need to develop this level of accuracy.
“When they encounter a narrow gap, they either lift their wings up vertically or tuck them in completely, minimizing their width to that of their torso,” he says.
Can it work for drones?
The researchers wanted to know precisely how birds judge gaps between obstacles before engaging in evasive maneuvers.
In testing, budgies flew down corridors with variable widths between obstacles, and their flights were recorded with high-speed cameras for analysis.
The research, published in the journal Frontiers in Zoology, will be applied to robotics work at the Queensland Brain Institute’s Neuroscience of Vision and Aerial Robotics laboratory, Schiffner says.
“If we can understand how birds avoid obstacles, we might be able to develop algorithms for aircraft to avoid obstacles as well.
Bird brains
“For instance, urban drones used for deliveries would need to fly through complex environments such as tight alleyways or between trees at the front of homes.
“For us, it isn’t the ability to tuck in wings that is of interest if we are talking about fixed-wing or rotor aircraft, but whether we can replicate what happens neurologically in birds as they navigate.”
To judge airspeed, budgies use optic flow—the rate visual cues pass by the eyes. They don’t see three-dimensionally like humans, due to the lateral placement of their eyes and lack of binocular overlap.
“Seeing in three dimensions requires two eyes or cameras with sufficient visual overlap, so using optic flow with just one camera would be very useful, saving weight and keeping autonomous vehicles small.”
Source: University of Queensland
