Who doesn’t enjoy watching hummingbirds through the window at their backyard feeder? These amazing birds, zipping to and fro in all directions at stunning velocity, never seem to crash. Intrigued by their flight prowess, Canadian scientists decided to look into how they do it. They learned something new and unique about the tiny birds’ strategy for collision-free navigation.

To make controlled observations of hummingbird paths, the team from the University of British Columbia built a flight simulator consisting of a long box 5.5 meters long, lined with eight cameras. On the side walls, they could project images of vertical and horizontal lines of various widths, and control their motions. Then they caught 18 wild hummingbirds, kept them separate from one another, and trained them to fly the tunnel toward a feeder. While projecting a wide variety of patterns on the walls, they filmed 3,100 flights by the birds.

The results are published in the Proceedings of the National Academy of Sciences (PNAS). New Scientist says:

Hummingbirds have a unique collision avoidance system built into their brains that allows them to perform high-speed aerobatics in safety.

The super-agile birds, whose wings beat up to 70 times a second, can hover, fly backwards, and whizz through dense vegetation at more than 50 kilometres per hour.

How they manage to avoid potentially fatal crashes has remained a mystery until now. Researchers in Canada conducted a series of experiments which showed that the birds process visual information differently from other animals. [Emphasis added.]

Here’s what the team expected. Based on earlier studies with insects, they thought that birds would respond to moving stripes on the side walls. The paper states:

Information about self-motion and obstacles in the environment is encoded by optic flow, the movement of images on the eye. Decades of research have revealed that flying insects control speed, altitude, and trajectory by a simple strategy of maintaining or balancing the translational velocity of images on the eyes, known as pattern velocity. It has been proposed that birds may use a similar algorithm but this hypothesis has not been tested directly.

To their surprise, the scientists found that hummingbirds use a different strategy. They don’t seem to be affected by pattern velocity, at least when seeing vertical stripes pass by along the walls (known as “nasal-to-temporal pattern velocity”). One would think that the speed of moving bars in their peripheral vision (i.e., the optic flow), would make them speed up or slow down as they approached the feeder; this is known as the “telegraph pole effect,” familiar to drivers who can gauge their speed by the passing of telephone poles. But no matter how they varied the speed of the vertical stripes, it didn’t seem to matter to the birds. New Scientist explains:

When the scientists simulated the “telegraph poles effect” with vertical moving stripes, the hummingbirds did not react. Instead, they appeared to rely on the size of objects to determine distance, steering away from the stripes as they grew larger.

“When objects grow in size, it can indicate how much time there is until they collide even without knowing the actual size of the object,” says Dakin. “Perhaps this strategy allows birds to more precisely avoid collisions over the very wide range of flight speeds they use.”

It’s not that the birds were incapable of responding to the speed of the moving bars. When horizontal bars were projected moving up or down, the birds did respond by varying their altitude, probably using the information to gauge their risk of hitting the ground. But for preventing collision with the feeder, they appeared to rely mainly on the increasing vertical size of the target.

Further tests confirmed their findings. Instead of stripes, they tried projecting moving dots. They tested many combinations of dots, stripes, stripe orientations and stripe motions. The secret became clear: expansion of an object in any part of the field of view was the bird’s cue to respond. They would slow down or steer away from anything that grew in size vertically.

Collectively, our findings suggest that birds control forward flight by monitoring changes in the vertical axis: specifically, the height of features and vertical pattern velocity. This finding is consistent with other laboratory studies showing that flying birds rapidly stabilize key features in their visual field. In nature, collisions may be avoided by monitoring changes in the apparent size of features, such as trees and branches, as well as changes in the vertical position of those features. Although our experiments focused on manipulating a limited number of cues, we do not suggest that these represent the only visual guidance strategies used by birds.

Now apply this to a confusing, dynamic tangle of branches, leaves, flowers, and objects that hummingbirds must face in their natural environments. The strategy allows them to quickly zero in on the pertinent optic flow information to avoid collisions. Moving at 50 km/hour, hummingbirds must be quick! This programmed strategy avoids TMI (too much information), giving them only what they need at the moment, even though their brains are perfectly capable of gathering and processing all the information in the visual field.

It makes sense that birds, being much larger than insects, would use a different strategy for collision avoidance. It appears that the strategy is common to birds. Previous tests with budgerigars (the pet parakeets) showed similar responses. Now see what a large goshawk has to deal with in its flight through a tangled forest!

The scientists realize that the behaviors must relate somehow to actual nerve impulses:

Neurons that compute expansion have been identified in the nucleus rotundus of the pigeon brain, part of the tectofugal pathway…. These cues can inform an animal about the nearness in time of an impending collision, triggering an appropriately timed response without knowledge of the true size or distance of the approaching object. It was recently discovered that the zebra finch nucleus rotundus also contains cells that respond during simulated flight if an approaching feature is located at the point of expansion, suggesting that the tectofugal pathway may also be involved in flight control.

Flight control. That’s design. Understandably, the scientists did not speculate about how flight control systems might have evolved. Spinning a “narrative gloss” on the findings would have dubious value.

On the contrary, you can certainly appreciate that the knowledge gained by investigation of “flight control” in hummingbirds might improve the design of drones, now that “drone racing” is becoming one of the hottest new sports. Look at this:

As you watch the man-made racing drones suffer “spectacular crashes” in their “daring aerial maneuvers,” then realize that hummingbirds already had collision avoidance figured out, you can’t help but remember Paul Nelson’s quip in Flight: The Genius of Birds, “If something works, it’s not happening by accident.”

Photo source: University of British Columbia.