Ravi Umadi M.Tech

TUM School of Life Sciences
Technische Universität München
Freising, Germany

Connect with me

I am a behavioural bioacoustician and experimentalist, fascinated by how animals interact with their environment through sound. My doctoral research at the Technical University of Munich focused on echolocation dynamics in bats, particularly how they adapt their sonar system during active foraging and pursuit. I combined theoretical modelling, laboratory experiments with real-time virtual acoustic environments, and field studies to explore how bats modulate sound production, ear movements, and flight behaviour to optimise prey detection and capture.

A key part of my work has been developing new experimental and computational methods to study these processes, including ways to model and simulate acoustic interactions with dynamic morphology—such as moving ears or noseleaves—that shape auditory perception. This has allowed me to uncover new insights into emitter-receiver coordination and the sensory strategies bats use in complex environments.

Beyond my core work on bats, I am deeply interested in applying generative simulation and AI-based modelling to broader questions in sensory biology. I enjoy building tools that bridge biology, physics, and computation, and I am constantly driven by the challenge of understanding how living systems exploit physical laws to navigate and survive in the world.

Selected Work

Doctoral dissertation

• 1.

  Umadi, Ravi (2025). Echolocation Dynamics in Active Foraging and Pursuit. TUM School of Life Sciences. Submitted.

  Abstract

  Echolocating bats rely on self-generated biosonar signals to perceive and navigate their environments. While traditional models treat echolocation as a rigid, reflexive behaviour with fixed emitter and receiver geometries, empirical evidence suggests a high degree of plasticity in both signal generation and spatial sampling. This thesis explores the dynamic sonar strategies employed by the omnivorous bat Phyllostomus discolor, focusing on how emitter-receiver decoupling, deformable noseleaf morphology, and context-sensitive signal timing enhance acoustic sensing flexibility. I first present measured beam geometric and a computational beam model that treats the bat's noseleaf as a two-point source rather than a conventional piston emitter. This approach reveals that small deformations in the noseleaf can result in substantial changes to the spatial profile of the sonar beam, including shifts in beam direction and width. My results show that such modulation is functionally significant, allowing the bat to alter its acoustic gaze without body reorientation. Subsequently, I examine the degree of coordination between emission and reception axes during fixed-ears and free-moving conditions. Using high-resolution motion capture and stereo microphone arrays, I demonstrate that P. discolor exhibits fast, independent pinna movements that enable acoustic sampling across a wide spatial field. This emitter-receiver decoupling enhances spatial resolution and increases the likelihood of detecting novel or peripheral targets, supporting a broader sensory field during foraging and navigation. In the final study, I analyse changes in sonar signal structure during target approach. I show that bats regulate their call rate and duration based not only on distance to target but also on task urgency and context variability. The terminal buzz, traditionally viewed as a motor constraint, is instead framed here as a responsive sensory strategy that modulates information update rate dynamically. Together, these findings reveal a sophisticated system of acoustic sensing grounded in morphological flexibility, sensorimotor coordination, and adaptive control. By integrating physical modelling, behavioural experiments, and signal analysis, my work contributes to a growing view of echolocation as an active and embodied perceptual process. These insights have implications for the evolution of biosonar systems and offer promising design principles for artificial sensing technologies, particularly in robotics and autonomous navigation.

Methods

• 1.

  Umadi, Ravi (2025). Embedded Ultrasonics - A Microcontroller-Based Multichannel Ultrasound Recorder for Behavioural Field Studies. Submitted. https://www.biorxiv.org/content/10.1101/2025.08.11.669530v1

  Abstract

  1. Behavioural studies investigating acoustic communication in animals—particularly echolocating bats—require transducers and recording systems that are lightweight, power-efficient, and easy to deploy. However, capturing high-frequency calls typically demands specialised, high-cost equipment and laptops, which limit portability and scalability in field settings. Advances in embedded microcontrollers and high-fidelity MEMS microphones offer an underexplored opportunity for building compact, affordable, and open-source recorders. These platforms provide design flexibility and a growing support community, but are seldom adapted for multi-channel ultrasonic recording in behavioural research. 2. To address this gap, I developed Batsy4-Pro, a 4-channel ultrasonic recorder based on the Teensy 4.1 microcontroller. The system uses WM8782 ADCs to acquire synchronised 192 kHz audio streams and records them to a microSD card. The firmware is written in C++ on Arduino IDE and allows flexible configuration of recording parameters, buffering schemes, and triggering logic. The system weighs under 150 g and runs on a 5 V DC power supply, enabling untethered field deployment. 3. The system also integrates real-time heterodyne monitoring using a PCM5102A DAC, allowing users to audibly monitor bat activity during field experiments, reducing the need for additional equipment and helping field researchers decide when to initiate recordings. 4. Using a custom-built microphone array, I validated its recording fidelity using both synthetic bat call playbacks under controlled conditions and free-flying bat vocalisations in a natural foraging corridor. Recordings consistently showed high signal-to-noise ratios (>40 dB), suitable for accurate call detection and spatial localisation. 5. Batsy4-Pro offers an accessible and extensible tool for ultrasonic recording in behavioural, ecological, and neuroethological research, supporting studies where portability, ease of use, and field-readiness are essential. The open-source code and modular design facilitate community-driven development and promote the system's adoption, enabling the evolution of experimental protocols and recording paradigms, as well as the exploration of novel approaches to behavioural experiments in natural settings.

• 2.

  Umadi, Ravi (2025). ESPERDYNE - A Dual-Band Heterodyne Monitor and Ultrasound Recorder for Bioacoustic Field Surveys. Submitted. https://www.biorxiv.org/content/10.1101/2025.08.30.673206v1

  Abstract

  Abstract. 1. Background. Ultrasonic monitoring is essential for ecological studies of bats and other animals, yet high‐performance field devices remain prohibitively expensive and inaccessible–particularly in biodiversity‐rich regions with limited research infrastructure. Existing low‐cost options often lack real‐time listening and require a complex setup. There remains a critical need for versatile, affordable, and field‐ready tools that support acoustic behavioural research, educational and conservation outreach. 2. New Tool. I introduce Esperdyne, an open‐source, dual‐channel ultrasound monitoring and recording system based on the ESP32‐S3 microcontroller. With a component cost under €75, Esperdyne combines real‐time heterodyne monitoring, stereo recording from a retroactive ring buffer, and an intuitive rotary‐based user interface with OLED display. It supports full‐duplex 192 kHz audio, dual‐band tuning for simultaneous FM/CF monitoring, and real‐time playback via headphones or a speaker. All audio processing–including adjustable carrier frequency mixing, gain control, and file‐saving logic–is implemented without reliance on fixed‐rate audio libraries. 3. Applications. Esperdyne has been tested in field conditions and shown to reliably detect high‐SNR calls and harmonics from free‐flying bats. A companion MATLAB tool Bat Reviewer supports rapid inspection, playback, and export of selected recordings. Together, these tools enable portable, solo‐operated acoustic surveys with minimal training. Beyond ecological research, Esperdyne is suitable for education, outreach, and preliminary field assessments in remote or resource‐constrained settings. Its modular design encourages hardware customisation and firmware extension by interdisciplinary teams. 4. Availability and Implementation. Full hardware schematics, firmware, and software tools are publicly available. The system can be built using hobbyist‐accessible components and standard Arduino tooling. By sharing this system openly, I aim to lower technical barriers and foster broader participation in ultrasound‐based biodiversity monitoring and conservation. Esperdyne demonstrates how microcontroller‐based platforms can bridge gaps between affordability, usability, and scientific capability–supporting global efforts in soundscape ecology.

Research

• 1.

  Umadi, Ravi (August 25, 2025). Widefield Acoustics Heuristic - Advancing Microphone Array Design for Accurate Spatial Tracking of Echolocating Bats. BMC Ecology and Evolution. https://doi.org/10.1186/s12862-025-02441-4

  Abstract

  Accurate three-dimensional localisation of ultrasonic bat calls is essential for advancing behavioural and ecological research. I present a comprehensive, open-source simulation framework—Array WAH —for designing, evaluating, and optimising microphone arrays tailored to bioacoustic tracking. The tool incorporates biologically realistic signal generation, frequency-dependent propagation, and advanced Time Difference of Arrival (TDoA) localisation algorithms, enabling precise quantification of both positional and angular accuracy. The framework supports both frequency-modulated (FM) and constant-frequency (CF) call types, the latter characteristic of Hipposiderid and Rhinolophid bats, which are particularly prone to localisation errors due to their long-duration emissions. A key innovation is the integration of source motion modelling during call emission, which introduces Doppler-based time warping and phase shifts across microphones—an important and often overlooked source of error in source localisation. I systematically compare four array geometries—a planar square, a pyramid, a tetrahedron, and an octahedron—across a volumetric spatial grid. The tetrahedral and octahedral configurations demonstrate superior localisation robustness, while planar arrays exhibit limited angular resolution. My simulations reveal that spatial resolution is fundamentally constrained by array geometry and the signal structure, with typical localisation error ranging between 5-10 cm at 0.5 m arm lengths. By providing a flexible, extensible, and user-friendly simulation environment, Array WAH supports task-specific design and deployment of compact, field-deployable localisation systems. It is especially valuable for investigating the acoustic behaviour of free-flying bats under naturalistic conditions, and complements emerging low-power multichannel ultrasonic recorders for field deployment and method validation.

• 2.

  Umadi, Ravi (2025). Swarm Cohesion in Bats Emerges from Stable Temporal Loops. Under Review. https://www.biorxiv.org/content/10.1101/2025.07.05.663265v1

  Abstract

  Swarming in echolocating bats presents a compelling example of decentralised coordination driven by acoustic sensing. Unlike visually guided animals, bats navigate and maintain cohesion in dense groups using self-generated sonar signals. I present a biologically grounded simulation framework in which agents operate asynchronously using a closed-loop control policy based on local echo delays. Building on the theory of biosonar responsivity and the deduction of temporal precision in bat echolocation, my model predicts a tight coupling between echo delay, call duration, and call rate, and demonstrates that local, echo-timed interactions suffice to generate stable, self-organised swarm behaviour. To test these predictions, I systematically varied initial velocity and responsivity coefficient, kr, across a simulation grid and analysed the resulting call dynamics and collision risks. An information propagation model was derived, estimating the effective spatiotemporal speed of behavioural updates across the swarm. I also developed a decay-based model of perturbation attenuation to quantify the spatial limit of influence for local disturbances. Together, these analyses reveal a trade-off between responsiveness and stability - lower kr yields faster information flow but higher collision risk in denser regions, while higher kr promotes stability but reduces reactivity. My results align closely with recent empirical findings and provide a generative explanation for swarm cohesion based on the temporal precision of echolocation. These insights offer a promising design framework for decentralised control in bioinspired robotic swarms, where stability, autonomy, and resilience emerge from local sensing and feedback, rather than explicit synchrony or global coordination.

• 3.

  Umadi, Ravi (2025). Temporal Precision Necessitates Wingbeat-Call Asynchrony in Actively Echolocating Bats. Under Review. https://doi.org/10.1101/2025.06.18.660328

  Abstract

  I present a unified theory and empirical analysis showing that temporal precision, rather than energetic efficiency or mechanical synchrony, is the primary axis guiding echolocation call timing in flying bats. While classic hypotheses posited that coupling call emission to wingbeat and respiration cycles is optimal, my field data and mathematical modelling demonstrate that such synchrony is only maintained when sensory-motor demands are minimal. As bats approach a target or encounter complex, dynamic environments, synchrony is frequently and necessarily broken(:) I show that the strict temporal constraints imposed by the call-echo-response loop require bats to decouple vocal output from wing motion whenever echo delays become short or as circumstantial demands for information updates dictate. Using a simulation framework grounded in first principles, I reveal that wingbeat-call synchrony is possible only within a narrow physiological window, bounded by wingbeat frequency and amplitude. When these limits are exceeded, asynchrony reliably emerges as the only viable strategy to maintain real-time sensory feedback. Both my empirical data and theoretical model predict and explain the universal emergence of a hyperbolic relationship between interpulse interval and call rate-across all behavioural and environmental contexts-demonstrating that closed-loop, echo-guided timing is a fundamental, conserved feature of bat biosonar. Patterns such as sonar sound groups arise not as discrete modules, but as visible signatures of the feedback-driven system flexibly adapting to heightened uncertainty or unpredictability. I further discuss how species-specific morphological and aerodynamic constraints set the boundaries for synchrony flexibility, explaining interspecific diversity in echolocation behaviour. Altogether, these findings demonstrate that wingbeat-call asynchrony is an adaptive, mathematically inevitable solution for temporal precision in active echolocation, unifying previously disparate empirical observations and providing a predictive foundation for future research in sensory-motor coordination, flight control, and biosonar.

• 4.

  Umadi, Ravi and Firzlaff, Uwe (2025). Biosonar Responsivity Sets the Stage for the Terminal Buzz. Under Review. https://doi.org/10.1101/2025.06.16.659925

  Abstract

  Echolocating bats must continuously adapt their sonar output to meet the increasing demands of prey pursuit. While call rate and duration modulations have been extensively described, the underlying control thresholds governing these transitions remain poorly understood. Here, we present a predictive framework based on a novel metric responsivity, defined as the inverse change in interpulse interval (IPI), to quantify moment-to-moment temporal precision in sonar control. This metric reveals a critical transition point---buzz readiness---where fine-scale IPI adjustments peak prior to the onset of the terminal buzz. By integrating biologically plausible reaction time constraints with echo-acoustic feedback loops, our model predicts how increasing relative velocity compresses the time available for sonar adaptation. Simulations incorporating prey motion and bat flight kinematics reproduce a consistent sublinear tradeoff between call rate and relative velocity. High-resolution field recordings from a portable, custom-built microphone array validate the model predictions, demonstrating that buzz readiness thresholds reliably align with behavioural transitions in natural prey interception. The framework further explains why shortened or absent buzzes occur at high velocities, when reaction constraints prevent full transition into the buzz phase. This work introduces a generalised, predictive model linking sensory-motor control, kinematics, and temporal adaptation in bat biosonar. The responsivity approach offers new tools to quantify control dynamics in natural behaviour and provides a foundation for biologically inspired sensing systems operating under real-time constraints.

• 5.

  Umadi, Ravi (2025). Oscillating Ears Dynamically Transform Echoes in Constant-Frequency Bats. In Revisions. https://doi.org/10.1101/2025.06.14.659613

  Abstract

  The oscillatory movements of the pinnae in constant-frequency (CF) bats have long been documented, yet their role in actively transforming echo signals has remained underexplored. Inspired by Perrine's foundational description of Doppler effects in oscillating receivers, I hypothesised that bat ear oscillations could serve as dynamic signal modulators, introducing time-varying Doppler shifts to received echoes. I tested this hypothesis using both computational simulations and controlled experiments. In the simulations, each ear was modelled as an angularly oscillating structure receiving echoes from a CF pulse, with motion parameters systematically varied across frequency (5-50 Hz) and excursion angle (15-45 degrees). The received echoes were synthesised as composite signals comprising delayed, Doppler-shifted components based on instantaneous ear segment velocities. Results showed pronounced spectral broadening, instantaneous frequency fluctuations, and binaural disparities that scaled with tip velocity. These effects were then validated using a subwoofer-driven oscillating reflector that emulated the ear's motion while reflecting a stable 80 kHz tone. Recordings showed consistent bandwidth expansion to over 1.2 kHz at 30 Hz, with phase warping evident in the time-domain waveforms. These findings confirm that oscillatory motion alone, without source modulation or target motion, can dynamically restructure the spectral and temporal profile of echoes. I argue that in CF bats, such echo transformations represent an active encoding strategy, enabling spatial and temporal contrast enhancement within the auditory fovea. This reconceptualises ear motion not as passive filtering, but as a means of dynamically enriching incoming sensory information.