Rolf Müller - firstname.lastname@example.org
John C. T. Hallam
The Maersk Institute, University of Southern Denmark
DK-5230 Odense M, Denmark
(University of Antwerp, Belgium)
Alexander Streicher, Reinhard Lerch
(University of Erlangen-Nürnberg, Germany)
Popular version of paper 4aAB6
Presented Thursday morning, May, 27th, 2004
147th ASA Meeting, New York, NY
Bats are flying mammals and as such share a common anatomical blueprint with other mammals and humans. However, certain features in this blueprint have been substantially modified in order to accommodate the special needs of their nocturnal lifestyles, in particular how they broadcast ultrasound and then detect its echoes to sense their environment (biosonar). The most obvious example of these modifications are the grotesque facial features which have inspired artists for centuries. Each of the approximately 1000 different kinds (species) of bat which occur worldwide has molded the common features of a mammal's ear into a highly optimized, sophisticated sonar antenna. Examples of the structural features used in this optimization are the "tragus" and the "antitragus." In humans, these are small, inconspicuous knobs at the frontal and back rim of the outer ear opening. Among the general public, knowledge of these structures by name is mostly limited to body-piercing enthusiasts, who use them as attachment points for body jewelry. In bats, tragus or antitragus are frequently grown into very conspicuous flaps which obviously influence waves passing through the outer ear opening.
We have developed a set of methods which can predict the acoustic properties of these natural biosonar antennae. Our method allows us, for the first time to our knowledge, to map the sensitivity of a bat's ear to low-, -mid, and high frequencies in three dimensions and with high resolution. Furthermore, we are able to link anatomical features to the properties of this spatial sensitivity pattern. Our methods take a biological sample (e.g., a bat ear) as an input and generate as output a three-dimensional view of how the ear probes space as the animal whistles ("chirps") through the different ultrasonic frequencies contained in its sonar pulses.
Renderings of shape representations reconstructed from computer-tomographic scans of bat ears (and one nose, shown in the center).
The first step necessary to achieve this is to obtain a digital representation of the shape of the ear or any of the other facial protrusions seen in bats. These shapes determine the interaction between facial structures and the emitted or received ultrasonic waves. The consequence of these interactions is a spatial distribution of sound energy or receiver sensitivity over the "sonar spotlight" which the animal shines around in the environment. An animation of a reconstructed ear shape can be seen here (1.7 MByte). The animation shows, how the ear shape is represented by a mesh of triangular elements, which is a standard way of shape representation in computer graphics.
Examples for simulated ultrasonic wavefields emerging from the ear. Left: sound amplitudes (related to loudness) coded by color; right: wave contours (animated version - 3.5 MByte).
In the next step, the interaction ("diffraction") between shape and ultrasonic waves has to be modelled. This is done by a computer calculation known as the "finite element method". It simulates a physical process by modelling space as a stack of tiny, discrete elements. Here, this modelling approach was used to simulate how the ears would act as a "loudspeaker" which emits sound waves. Although this is not their natural function, their spatial sensitivity as a "microphone" can nevertheless be inferred from the results. Simulating a loudspeaker instead of a microphone was used here as a procedural trick to speed up the simulations. An animation of a simulated wave emerging from a bat outer ear can be seen here (3.5 MByte).
Prediction of the spatial sensitivity of a bat ear. Sensitivity isosurfaces are rendered in a different color for each frequency (blue colors: low frequencies, red colors: high frequencies). An animated version is available here (4.2 MByte).
In the final step, the simulated ultrasonic waves are used to predict the spatial sensitivity of the ear. For each location in space, an estimate is made of how sensitive the ear would be for an ultrasonic source (or a sound-reflecting target) at this particular location. This sensitivity depends on the frequency of the sound: at certain frequencies, the ear may be very sensitive in a particular spot, whereas at others, it may almost be deaf in this spot and sensitive somewhere else. In order to comprehend, how the sensitivity of an ear is distributed in space for different frequencies, one sensitivity isosurface is rendered for each frequency using a different color to mark the frequency. In the image, isosurfaces belonging to low frequencies are shown in blue colors, those belonging to high frequencies in red colors. In animations like this one (4.2 MByte), it is possible to see how the bat ear "looks around" in space as the sonar pulse whistles through different ultrasonic frequencies.
Examples of shape manipulations. Left: Boolean surgery; center: rotation of the tragus, right: digital face(ear)-lift.
The scientific approach for any study of physical causes and effects is to manipulate the cause and look for a disruption of the effect. Similarly, in order to understand the physical causes behind an ear's spatial sensitivity, it is necessary to modify the ear shapes experimentally. Since a digital representation of the shape is available here, the powerful tools of computer graphics can be used for this purpose. The ear shapes can be subjected to the same kinds of shape change which characters in computer animated movies frequently undergo. For example, structural features like the tragus or antitragus can be removed - painlessly - by Boolean surgery or systematically rotated by an appropriate transform. Surface ripple and other ridge patterns can be removed by digital face- or in this case ear-lift. By comparing the spatial sensitivities of the manipulated shapes with the original shapes and among each other, hypotheses for causal links between structure and function can be formulated.
Studying a collection of approximately 1000 smart antenna designs, which have been perfected in at least 60 million years of evolutionary optimization, taps into a knowledge base from which technologically relevant insights may be drawn. Antennas are essential components of many technical devices, used in application areas such as medical diagnosis (based on ultrasound or otherwise), wireless communication, and surveillance. Novel, smart ways of distributing the emitted signal or receiving sensitivity of an antenna in space with frequency have the potential to improve these devices. Such improvements could, for example, extend the diagnostic and sensing capabilities or improve the quality of the available communication links.
This work is supported by the European Union (CIRCE Project, IST-2001-35144).