Biosonar Task Research Articles

Investigating the impact of biomimetic pinna shape variations on clutter echoes received from natural environments

Bat species navigating dense vegetation based on biosonar have to obtain the necessary sensory information from “clutter echoes”, i.e., echoes that are superpositions of contributions from many reflecting facets (e.g., leaves) and hence have highly unpredictable waveforms. Prior results have suggested that pinna motions could aid in direction-finding tasks based on deterministic echo patterns. This raises the question whether varying pinna shapes could also have a function significance for challenging biosonar tasks performed on clutter echoes. As a first, task-independent step to test this hypothesis it has been investigated whether different pinna shapes have a consistent effect on clutter echoes despite the random nature of these signals. This was accomplished using a dedicated laboratory setup that produced large amounts of uncorrelated clutter echo data by agitating an artificial foliage with fans between echoes. Deep learning methods were then applied to identify the pinna shape that received a given clutter echo. A two-dimensional convolutional neural network operating on spectrograms achieved 90% validation accuracy in this task. This finding demonstrates that even small pinna deformations can impart consistent effects on the clutter echoes. Ongoing research is directed at analyzing the nature of the signal properties that the successful classifications were based on.

A mammal for all seasons: Contributions from the bottlenose dolphin TRO to bioacoustics research at the US Navy Marine Mammal Program

Among the notoriously small sample sizes of marine mammals in cooperative behavioral research, there is an even smaller subset of subjects that can be considered “experts.” In essence, these are subjects that have “learned how to learn” in the contexts of novel experimental paradigms. One such subject is the bottlenose dolphin TRO at the US Navy Marine Mammal Program. The first scientific publication including TRO heralded his potential; aided by a phantom echo generator, he was likely the first marine mammal in history to echolocate while out of water. Subsequent behavioral experiments with TRO have examined numerous aspects of biosonar, including transmit beam characteristics and the use of click “packets” during long-range target detection. The patient nature of TRO has made him an ideal subject for electrophysiological studies, where he has diligently listened to exotic acoustic stimuli designed to examine auditory processing in the dolphin brainstem. The most significant scientific contributions from TRO are arguably from studies that have used auditory evoked potential methods during biosonar tasks. These data provide the temporal resolution necessary to examine hearing at the level of click production and echo reception and are a result of TRO’s dependable participation in studies combining multiple experimental techniques

Autonomy, soft-robotics, deep learning, and bat biosonar

Echolocating bat species that inhabit dense vegetation are promising model systems for achieving autonomy in complex natural environments. To replicate the skills of the bats in extracting information about complex environment that can support autonomous navigation in a man-made system three aspects deserve particular attention: (i) encoding of relevant sensory information; (ii) information extraction; and (iii) integration of information encoding and extraction with each other and the respective context. To facilitate progress towards understanding these issues, a biomimetic sonar system is being developed that is aimed at recreating the flexibility and variability that bats exhibit in the configuration of the emission and reception baffles of their biosonar systems (i.e., noseleaves and pinnae). Analyzing the significance of these dynamic effects on sensory information encoding requires an understanding of the stimulus ensemble of biosonar tasks in natural environment. To this end, the biomimetic sonar system has been used to collect large data sets with echoes that pertain to fundamental navigation tasks such as localization and passageway finding. Deep-learning methods have been shown to be a good match for analyzing this echo data. Finally, ongoing research is directed at data acquisition and inference with the specifics of a given habitat and biosonar task.

Detection of simulated patterned echo packets by bottlenose dolphins (Tursiops truncatus)

Dolphins performing long-range biosonar tasks sometimes use "packets" of clicks, where inter-click-intervals within each packet are less than the two-way acoustic travel time from dolphin to target. The multi-echo nature of packets results in lower detection thresholds than single echoes; however, other potential benefits of packet use remain unexplored. The present study investigated whether structured temporal patterns observed in click packets impart some advantage in detecting echo-like signals embedded in noise. Two bottlenose dolphins were trained to passively listen and detect simulated packets of echoes in background noise consisting of either steady-state broadband Gaussian noise, or Gaussian noise containing randomly presented impulses similar to dolphin clicks. Four different inter-stimulus-interval (ISI) patterns (constant, random, increasing, or decreasing ISI within each packet) were tested. It was hypothesized that decreasing ISIs-found naturally in dolphin packets-would result in the lowest thresholds, while random, unlearnable patterns would result in the highest. However, no biologically significant differences in threshold were found among the four ISI patterns for either noise condition. Thus, the bottlenose dolphin's stereotypical pattern of decreasing ISI during active echolocation did not appear to provide an advantage in packet detection in this passive listening task.

Dolphin echo-delay resolution measured with a jittered-echo paradigm.

Biosonar echo delay resolution was investigated in four bottlenose dolphins (Tursiops truncatus) using a "jittered" echo paradigm, where dolphins discriminated between electronic echoes with fixed delay and those whose delay alternated (jittered) on successive presentations. The dolphins performed an echo-change detection task and produced a conditioned acoustic response when detecting a change from non-jittering echoes to jittering echoes. Jitter delay values ranged from 0 to 20 μs. A passive listening task was also conducted, where dolphins listened to simulated echoes and produced a conditioned acoustic response when signals changed from non-jittering to jittering. Results of the biosonar task showed a mean jitter delay threshold of 1.3 μs and secondary peaks in error functions suggestive of the click autocorrelation function. When echoes were jittered in polarity and delay, error functions shifted by approximately 5 μs and all dolphins discriminated echoes that jittered only in polarity. Results were qualitatively similar to those from big brown bats (Eptesicus fuscus) and indicate that the dolphin biosonar range estimator is sensitive to echo phase information. Results of the passive listening task suggested that the dolphins could not passively detect changes in timing and polarity of simulated echoes.

Nearfield and farfield measurements of dolphin echolocation beam patterns: No evidence of focusing.

The potential for bottlenose dolphins to actively focus their biosonar transmissions was examined by measuring emitted clicks in four dolphins using horizontal, planar hydrophone arrays. Two hydrophone configurations were used: a rectangular array with hydrophones 0.2 to 2 m from the dolphins and a polar array with hydrophones 0.5 to 5 m from the dolphins. The biosonar task was a target change detection utilizing physical targets at ranges from 1.3 to 6.3 m with all subjects and "phantom" targets at simulated ranges from 2.5 to 20 m with two subjects. To provide a basis for evaluating the experimental data, sound fields radiated from flat and focused circular pistons were mathematically simulated using transient excitation functions similar to dolphin clicks. The array measurements showed no evidence that the dolphins adaptively focused their click emissions; axial amplitudes and iso-amplitude contours matched the pattern of the simulation results for flat transducers and showed a single region of maximum amplitude, beyond which spherical spreading loss was approximated.

Aural localization of silent objects by active human biosonar: neural representations of virtual echo-acoustic space.

Some blind humans have developed the remarkable ability to detect and localize objects through the auditory analysis of self-generated tongue clicks. These echolocation experts show a corresponding increase in 'visual' cortex activity when listening to echo-acoustic sounds. Echolocation in real-life settings involves multiple reflections as well as active sound production, neither of which has been systematically addressed. We developed a virtualization technique that allows participants to actively perform such biosonar tasks in virtual echo-acoustic space during magnetic resonance imaging (MRI). Tongue clicks, emitted in the MRI scanner, are picked up by a microphone, convolved in real time with the binaural impulse responses of a virtual space, and presented via headphones as virtual echoes. In this manner, we investigated the brain activity during active echo-acoustic localization tasks. Our data show that, in blind echolocation experts, activations in the calcarine cortex are dramatically enhanced when a single reflector is introduced into otherwise anechoic virtual space. A pattern-classification analysis revealed that, in the blind, calcarine cortex activation patterns could discriminate left-side from right-side reflectors. This was found in both blind experts, but the effect was significant for only one of them. In sighted controls, 'visual' cortex activations were insignificant, but activation patterns in the planum temporale were sufficient to discriminate left-side from right-side reflectors. Our data suggest that blind and echolocation-trained, sighted subjects may recruit different neural substrates for the same active-echolocation task.

Why dolphin biosonar performs so well in spite of mediocre ‘equipment’

Dolphins have been found to have an excellent sonar system that is able to detect and recognise targets in noisy and highly reverberant environments. However, their `equipment` has only mediocre characteristics from a technological sonar perspective. How dolphins perform the biosonar task so well is addressed in this manuscript. Echolocating dolphins have the capability to make fine discrimination of target properties such as wall thickness difference of water-filled cylinders and material differences in metallic plates, and to discriminate and recognise species of fish food. The high temporal resolution of the biosonar signals along with the high dynamic range of its auditory system are critical factors for target discrimination. An experiment in metallic plate composition discrimination suggests that dolphins attended to echoes 20`30`dB below the maximum level. Some of the properties of the dolphin sonar system are fairly mediocre, yet the total performance of the system is often outstanding. When compared to some technological sonar, the energy content of the dolphin sonar signal is not very high, the transmission and receiving beamwidths are fairly large, and the auditory filters are not very narrow. Yet the dolphin sonar has demonstrated excellent capabilities in spite of some mediocre features of its `hardware`.

Non-rigid pinna deformations in bats—Behavioral adaptation or dynamic sensing?

Mammal groups with exceptional spatial hearing abilities, bats and primates in particular, are known to have intricate pinna shapes. These shapes determine the monaural spatial hearing sensitivity as a function of direction and frequency, i.e., the reception beampattern. Since even comparatively small, local pinna features can have profound effects on the beampattern, it is possible that some of the interspecific diversity in these pinnae represents spatial hearing adaptations on an evolutionary time scale. In addition, some bat groups, such as horseshoe bats, have elaborate muscular actuation mechanisms that can produce non-rigid pinna deformations. These mechanisms could enable beampattern control on a much shorter, behavioral time scale. This could add considerably to the capabilities of the biosonar systems in these species, because by switching between different pinna shapes—and hence reception beampatterns—could meet even conflicting demands of different biosonar tasks. However, since pinna deformations in horseshoe bats can happen on comparatively short time scales (i.e., one tenth of a second), it is also conceivable that the bats' pinnae are in motion during the reception of an echo. It is not clear at present if and how such a dynamic receiver paradigm could be utilized to enhance biosonar performance.

Pinna-rim skin folds narrow the sonar beam in the lesser false vampire bat (Megaderma spasma)

False vampire bats (genus Megaderma) employ active as well as passive biosonars. In the present work, the acoustic impact of a conspicuous feature of the animals' ear morphology, skin folds of the pinna rim linking the two pinnae at the midline, has been studied using a numerical approach. Automated methods have been devised to measure the largest width of the beam patterns irrespective of beam orientation. A total of six pinna shapes from three individuals have been studied. For all these shapes, it was found that the reception biosonar beams had approximately elliptic cross-sections with the largest beamwidth being on average almost twice as large as the beamwidth in the orthogonal direction. The directions of the largest beamwidths were scattered around the azimuthal dimension. Removal of the skin folds resulted in significant widening of the beams along their widest dimensions with an increase in beamwidth of 9.2 degrees (a 30% change) on average. The strength and repeatability of this effect across individuals suggest the hypothesis that skin folds are functionally relevant to the animals' biosonar system. It may be a morphological adaptation to biosonar tasks that benefit from a narrow beam such as the detection of faint sounds or precise localization.

Chiroptera-inspired robotic cephaloid (CIRCE): A next generation biomimetic sonar head

The dynamic interplay between signal processing in the acoustic and neural domain renders the biosonar system of bats, a prime example of intelligent, integrated systems in biology. Previous attempts at technical reproductions of this system have studied a few functional aspects but did not produce convincing replicas of the integrated system. In particular, none of them could provide for beamforming pinnae, ear mobility, signal bandwidth and an adequate number of elements in the primary signal representation at the scale of the biological paragon. The CIRCE (http://www.circe-project.org) project brings together a multidisciplinary consortium with all competences necessary for addressing these aspects in conjunction. Well-integrated solutions to the following technological key challenges are therefore sought: (a) Evolution of pinna shapes to optimize performance in natural biosonar tasks. (b) Adaptation of state-of-the-art ultrasonic transducer technology for matching bats in terms of bandwidth and sound pressure level/sensitivity. (c) Actuation of mobile pinnae (including the possibility of nonrigid transformations) within the scale constraints of a bat’s head. (d) Custom-designed DSP hardware providing a qualitative and quantitative reproduction of the neural signal representation at the level of the auditory nerve. [Work supported by the European Commission, LPS Initiative.]

Interrelationships between intranarial pressure and biosonar clicks in the bottlenose dolphin (Tursiops truncatus)

Three Atlantic bottlenose dolphins (Tursiops truncatus) were given a target recognition biosonar task. During their performance of the task, both acoustic data in the far field and pressure within the bony nasal passages were digitally recorded (Elsberry et al., 1999). Analysis of over 15<th>000 biosonar clicks provided new insights into odontocete biosonar sound production and is consistent with acoustic and pressure data taken from white whales (Delphinapterus leucas) during biosonar (Ridgway and Carder, 1988). Our work provides the first evidence for a minimum intranarial pressure during biosonar click production for any odontocete (11.8±0.5 kPascals over basal pressure). All three subjects exhibited nearly the same minimum tranarial pressure difference during biosonar click production. Clicks produced at or near this minimum intranarial pressure exhibited a wide range of acoustic power values. The acoustic power of a biosonar click was not highly correlated with intranarial pressure (R2=0.116). The radiated acoustic energy in biosonar clicks ranged from 1 to 1370 microJoules. Estimates of mechanical work during pressurization events were produced using a piston/cylinder model and intranarial volume data from prepared specimens and computed tomography scans. Mechanical work during pressurization events ranged from 2.74 to 23.0 Joules, with an average of 10.3 Joules.