Chasing Echoes

This piece appeared in Imagine Magazine’s Nov/Dec 2007 issue. Imagine Magazine is published by the Center for Talented Youth at Johns Hopkins University.
You may reproduce this article only with proper attribution as follows:
Kaushik Ghose, “Chasing Echoes”, Imagine Magazine Nov/Dec 2007 issue.

I stand in a dark room holding a push-button switch, peering into a small video monitor. Something is clicking in the room. I glance at the oscilloscope next to me: the trace is flat. Another click, and the trace turns into a complex waveform with many cycles before going flat again. I glance up at a row of LEDs, which blink each time there is a click. Satisfied my equipment is working, I turn back to the video monitor and push the button on the switch.
A white pulsating form—a bat—makes a brief appearance on the video monitor and then disappears. The clicking gets louder, then softer. An insect appears, a white blob on the screen, whirring like a small fan. The clicking gets faster. A lot faster. The insect starts to weave as it crosses the screen. The bat reappears, now like the V silhouette of a bird, then turns sharply and dives after the insect. The clicking is now so fast, it’s a continuous buzz. The bat dives onto the insect, somersaulting. The buzzing stops. There is a faint rustling, like a starched dress. The bat changes into a ball, then back into a flapping V, and then leaves the screen. Alone.

“Did you get it?” someone asks from the darkness. We have spent most of the summer trying to make this eight-second video. I press the button once more. “I just did.”

Hearing the Way

Close your eyes and listen for a minute. Unless you are wearing headphones or in some strange environment, a universe of sound will creep into your awareness. Right now, I can hear my colleague working on his computer. With a little practice, I can probably tell whether he’s analyzing data, writing a paper, or reading e-mail. I hear footsteps in the corridor outside. I can identify several people from their footsteps and other sounds they make when they walk. From the sound of a human voice, we can guess, usually accurately, the gender, age, and physical size of the speaker. Sound carries a lot of information that we use, but are often not aware of.

Humans, like many other animals, normally rely heavily on sight to get around. If you know someone who is blind, you are most likely in constant awe of him or her. Blind people dress, make food, play soccer, go shopping, cross roads, and do target archery, among countless other activities. Some tasks might require a sighted guide or a specific organization, such as clothing arranged in stacks by color. Many spatial tasks, however, such as walking down a crowded sidewalk or playing soccer (with a special ball that emits a sound) are done without a guide. For such spatial tasks, blind people mainly use their hearing.

I am amazed at what people can do using only their hearing. But I am even more amazed by animals that fly in darkness, dodging through trees, performing split-second aerial maneuvers as they chase and capture tiny insects, guided only by sound. Bats, the only mammals capable of powered flight, have solved the problem of navigating and hunting in the dark by evolving an amazing biological version of sonar called echolocation.

Echolocating bats produce short chirps of high-pitched sound as they fly around. After making a chirp, the bat listens for the barrage of echoes that return from its surroundings. How the echoes sound tells the bat what objects are around it and where they are. They use this information to fly around at breakneck speeds, performing aerial acrobatics in three dimensions, chasing insects less than a centimeter in size. This is the part that amazes me: How can you tell where things are by how they sound?

Sound Check

Sounds have many qualities: a frequency structure, an envelope (how its loudness varies with time), and duration. The bat produces chirps. The objects around a bat, such as the ground below it, trees around it, and insects flying about it, all return echoes of this chirp. Each object, however, does not return a perfect echo. Each object changes the qualities of the sound it returns. How it changes the qualities of the sound depends on where the object is, what it’s made of, and what its size and shape is.

Perhaps the ground sounds “dull,” trees sound “echoey,” and insects sound “flinty.” Perhaps things in front sound “full,” while things off to the side sound “bass.” The bat’s brain converts this quality of sound into spatial information: “hill to the left, tree to the right, crunchy insect flying two o’clock low. Dive! Watch the branches from that tree!”

Just imagine that! Flying, not walking, on a dark night, weaving past scraggly branches and over undulating hills, looping and breaking as you chase tiny insects, with your eyes closed, making clicks with your mouth and listening for echoes.

As the huge video files from our high-speed infrared video cameras download onto disk, I walk around shutting off the apparatus. The bat’s pulses are ultrasonic, and my ears can only hear the very last bit, sometimes. We use ultrasonic microphones to record the sounds and display the waveform on an oscilloscope. We use a bat detector—a device that records the bat’s calls and converts them into audible clicks—to listen to the bat. The bat detector is now clicking occasionally; my bat has landed on the wall somewhere in the lab flight room and is “watching” me with his sonar system.

I switch off the array of microphones I use to record the bat’s calls. Each microphone is hooked up to an LED that flashes when it detects a bat sound. The microphone array enables me to compute which way the bat is scanning with its sonar beam as it flies around. Later, I will use the video to compute the flight path of the bat and of the insect it captured. I will combine information from the microphone array and the video to produce three-dimensional plots of where the bat was “looking” as it flew around the room and spotted and chased the flying insect.

Studies in Sonar

After studying how real bats fly and hunt insects for my doctoral degree, I decided to make a foray into creating “fake” bats. One way to understand something is to make a model of it. I’m currently working on creating computational models of bat sonar, trying to understand how neurons in the bat’s brain compute position from sound.

I’m particularly interested in the concept of “flow.” As you run down a street or chase a soccer ball, the visual scene flows by you. This flow gives you important information about where you are headed and how to maneuver through the environment. I suspect that bats also get information about how their surroundings flow past them, but the bat’s system is different from ours in several ways. The bat gets its information in discrete packets of information: one packet for each pulse it produces. Rather than a continuous stream of visual information, bats get snapshots of sonar signatures from their surroundings. How do bats stitch these discrete snapshots into a continuous whole that guides their rapid flight? By understanding how bat sonar works, we will uncover many general principles that apply to the process of using sensory information to guide movement (called sensorimotor integration). When you reach out to grab a cup that you see, your brain has to do a lot of complex computations to guide your hand to the spot your eye marks. Similarly, the bat’s brain has to do a lot of complex computations to guide its body to an insect that its sonar “sees.” Studying sensorimotor integration in a bat can reveal not only how a bat brain computes, but how other brains may compute.

One application of this research might be the construction of lightweight robots that operate using sonar. My fantasy project is a robotic lander that navigates the seas beneath the ice of Jupiter’s moon Europa. Vision and radar won’t work in that environment, and the probe needs to be autonomous. Studying the bat may be an excellent way to learn how to design an autonomous sonar-guided robot that flies through the seas of Europa.

But’s that in the future. Right now, I’m hoping that building computational models will help us understand a bit more about how bats perform their amazing and mysterious feat, catching echoes in the air.

Kaushik Ghose earned his PhD in 2006 from the Neuroscience and Cognitive Science program at the University of Maryland. He likes to go camping, do target archery with a recurve bow, take photographs, and sometimes just sit on the beach and read. Someday he will learn enough Bengali and German to write a book in those languages. Someday.

(c) Kaushik Ghose 2007