One big problem with putting autonomous drones to work delivering packages—or flying search-and-rescue missions—is that the sky is complicated and unpredictable. Trees, utility wires and spiraling footballs can turn up almost anywhere in the flight path. A new strategy for dodging these obstacles could come from an unexpected source: the way bats wiggle their ears.

The idea first occurred to Rolf Mueller, a Virginia Tech mechanical engineering professor, a few years ago while looking at bat photographs. He noticed that the ears of some species often looked blurry, because the animals were continually making rapid ear movements. But why?

Mueller studies bat behaviors, including their adaptations to the Doppler effect or Doppler shift. Both terms refer to the way sound waves from a fast-moving object such as a train or an ambulance get compressed—and therefore higher in pitch—as the object approaches a listener. Then the sound waves lengthen out again and become lower in pitch as the vehicle moves away. Even when the train or ambulance is out of sight, a person can tell roughly where it is at any moment from these changes in sound. Bats use the Doppler shift to locate objects in much the same way, but far more precisely.

Scientists have known since the 1930s that insect-hunting bats produce bursts of sound as they bob and weave through the night. They use the reflected sound waves to identify obstacles and target prey, an ability called echolocation or biosonar. Research in the 1960s showed that bats also interpret Doppler shifts, in sounds bounced off of flying insects, to zero in on a meal with high precision—even while maneuvering at breakneck speed through dense vegetation. More impressively, they can do this even though their flight movements produce their own potentially confusing Doppler shifts in the sound waves echoing back at them. The bats don’t get disoriented, because they adjust the frequency of the calls they produce to compensate for this “bad” Doppler shift.

“Since these groundbreaking discoveries, the general belief in the scientific community has been that the role of Doppler shifts in the biosonar systems of these animals has been completely understood,” Mueller says. But looking at photographs of blurry-eared bats, he wondered how rapid ear movements might fit into the complicated picture. He and co-author Xiaoyan Yin, a doctoral student in his lab, pursued two methods of testing whether the ears might actually move fast enough to produce Doppler shifts of their own—and what that might mean for echolocation. Their results were published this month in the Proceedings of the National Academy of Sciences USA.

First, working with trained horseshoe and leaf-nosed bats, the researchers painted visible spots at 60 points on each ear (concentrated on the more flexible outer ear). Then, using a synchronized array of high-speed cameras and ultrasonic microphones, they filmed each bat hanging from a perch as a novel object moved past. This demonstrated that the ear movements are fast enough to produce Doppler shifts, that these shifts are within a range bats can easily perceive, and that the shifts are timed to incoming echoes. Moreover, each ear moves independently and various parts of each deform at different times—“so at every point of time the bat is listening with a different ear shape,” Mueller says. He and Yin interpret this to mean that the bats are producing “good” Doppler shifts to alter the incoming sounds, thereby gaining more accurate information about the direction a biosonar target is traveling. To further test the idea the two researchers developed an artificial horseshoe bat ear out of silicon, with devices called “fast actuators” that move different parts of the ear in the same way bats do. These movements also added Doppler shifts to incoming sounds.

So what does any of this have to do with drones? Even the state-of-the-art versions being developed are “all pretty much clunkers when you compare them to bats,” Mueller says, especially if you “want the drone to be able to go into dense vegetation” (to do forestry work or search-and-rescue missions, for example). He and Yin have already designed what he calls a “bat hat” for a drone—a device that emits ultrasonic signals and has two moveable “ears” to record the echolocation. So far they have experimented with moving the device through a forest only by hand or on a zipline. “My career dream,” Mueller says, “is to have a drone that has the same agility in a natural environment as a bat.”

“I think that’s very smart,” says Melville J. Wohlgemuth, a neurobiologist at Johns Hopkins University, whose 2016 study of big brown bats was the first to show that bats use head waggles and ear movements as parts of their hunting process. “Artificial sonar and radar systems, and these autonomous systems for self-driving vehicles, could definitely be refined through some of the things we are learning from the biological world.” Research has shown, he adds, that “when we sense the world, we do it very actively. We’re constantly moving our eyes, and it’s the movement that enables us to take in information at high resolution.” Much the same thing now appears to be true for the way bats use their ears. Wohlgemuth, who was not involved in the new study, praises its findings as “highly plausible,” in part because of Mueller’s dual background in biological systems and engineering. “He walks that line very well,” Wohlgemuth says.

“When I first heard about this paper, I thought, ‘I don’t know,’” says Roman Kuc, a Yale electrical engineering professor whose work also walks that line, and who was not involved in the new research. “What they came up with—ear movements causing Doppler shifts—isn’t something I would have expected. Why is that believable?” Sensory mechanisms in the biological world can often seem counterintuitive, he suggests. “Engineers will tell you they want to maximize the signal-to-noise ratio. Bats do the opposite. They move theirs ears to the sides, which means sounds from the front are reduced. They reduce the signal-to-noise, but gain in the ability to echolocate.” Rapid ear movements now appear to be another such counterintuitive technique: the constant signal a bat emits is good for detecting the presence of an insect, Kuc says, but not necessarily for precisely locating it. But adding the Doppler shift as the sound returns to the ear makes it possible for the bat to track the insect’s location with pinpoint accuracy as it moves past, and even as it attempts to evade the bat. “This article,” Kuc says, “opens up a rather interesting complexity in the processing of echoes.”

One possible result: Doppler-shifting drones that can dodge airborne hazards to arrive safely at every doorstep.