arXiv subjects

Stefan Goetz

Publications and source records attributed to Stefan Goetz.

At least 19 recordsLinked to original sources

arXiv2026

SAM 3D Animal: Promptable Animal 3D Reconstruction from Images in the Wild

3D animal reconstruction in the wild remains challenging due to large species variation, frequent occlusions, and the prevalence of multi-animal scenes, while existing methods predominantly focus on single-animal settings. We present SAM 3D Animal, the first promptable framework for multi-animal 3D reconstruction from a single image. Built on the SMAL+ parametric animal model, our method jointly reconstructs multiple instances and supports flexible prompts in the form of keypoints and masks which enable more reliable disambiguation in crowded and occluded scenes. To train such a model, we further introduce Herd3D, a multi-animal 3D dataset containing over 5K images, designed to increase diversity in species, interactions, and occlusion patterns. Experiments on the Animal3D, APTv2, and Animal Kingdom datasets show that our framework achieves state-of-the-art results over both existing model-based and model-free methods, demonstrating a scalable and effective solution for prompt-driven animal 3D reconstruction in the wild.

cs.CV↗

arXiv2026

Optical Tag-Based Neuronavigation and Augmentation System for Non-Invasive Brain Stimulation

Accurate neuronavigation is critical for effective transcranial magnetic stimulation (TMS), as stimulation outcomes depend directly on precise coil placement. Existing neuronavigation systems are often costly, complex, and prone to tracking errors. To address these limitations, we present a computer vision based neuronavigation system that enables real time tracking of the patient and TMS instrumentation. The system integrates a multi camera optical tracking setup with consumer grade hardware and visible markers to drive a digital twin of the stimulation process. A dynamic 3D brain model in Unity updates in real time to visualize coil position and estimated stimulation targets. Augmented reality (AR) is further incorporated to project this model directly onto the patient's head, enabling intuitive, in situ coil adjustment without reliance on abstract numerical displays. Overall, the proposed approach improves spatial precision and accuracy while enhancing usability.

cs.HC↗

arXiv2026

Driver-Intention Prediction with Deep Learning: Real-Time Brain-to-Vehicle Communication

Brain-computer interfaces (BCIs) allow direct communication between the brain and electronics without the need for speech or physical movement. Such interfaces can be particularly beneficial in applications requiring rapid response times, such as driving, where a vehicle's advanced driving assistance systems could benefit from immediate understanding of a driver's intentions. This study presents a novel method for predicting a driver's intention to steer using electroencephalography (EEG) signals through deep learning. A driving simulator created a controlled environment in which participants imagined controlling a vehicle during various driving scenarios, including left and right turns, as well as straight driving. A convolutional neural network (CNN) classified the detected EEG data with minimal pre-processing. Our model achieved an accuracy of 83.7% in distinguishing between the three steering intentions and demonstrated the ability of CNNs to process raw EEG data effectively. The classification accuracy was highest for right-turn segments, which suggests a potential spatial bias in brain activity. This study lays the foundation for more intuitive brain-to-vehicle communication systems.

cs.HC↗

arXiv2026

Open-Source Coil Matching Toolbox for Magnetic Stimulation and Other Electromagnetics (COMATOSE)

The coil in transcranial magnetic stimulation (TMS) determines the spatial shape of the electromagnetic field in the head, which structures are concurrently activated, and how focal stimulation is. Most of the readily available coils have been designed intuitively instead of systematic mathematical-physical optimization as there were no methods available at the time. Previous research however demonstrated that these coils are far from optimum, e.g., for pulse energy or efficiency, and leave substantial room for lots of improvements. Techniques for rigorous mathematical optimization have been developed but are only available to very few groups worldwide. This paper presents an open-source toolbox, COMATOSE, to change that situation and make these methods available to a wider community. It incorporates the fundamental formalisms and offers vector space decomposition as well as base mapping as an explicit forward method, which is computationally less demanding than iterative computational optimization but can also form the initial solution for a subsequent optimization run if desired.

physics.med-ph↗

arXiv2025

Partial-Power Flow Controller, Voltage Regulator, and Energy Router for Hybrid AC-DC Grids

The share of electronically converted power from renewable sources, loads, and storage is continuously growing in the low- and medium-voltage grids. These sources and loads typically rectify the grid AC to DC, e.g., for a DC link, so that a DC grid could eliminate hardware and losses of these conversion stages. However, extended DC grids lack the stabilizing nature of AC impedances so that the voltage is more fragile and power flows may need active control, particularly if redundancy as known from AC, such as rings and meshing, is desired. Furthermore, a DC infrastructure will not replace but will need to interface with the existing AC grid. This paper presents a partial-power energy router architecture that can interface multiple AC and DC lines to enable precise control of voltages and both active as well as reactive power flows. The proposed system uses modular low-voltage high-current series modules supplied through dual active bridges. These modules only need to process a small share of the voltage to control large power flows. The topology reduces component size, cost, energy losses, and reliability more than three times compared to conventional technology. The optional integration of battery energy storage can furthermore eliminate the need for the sum of the power flows of all inputs to be zero at all times. Through dynamic voltage injection relative to the line voltage, the modules effectively balance feeder currents, regulate reactive power, and improve the power factor in AC grids. Real-time hardware-in-the-loop and prototype measurements validate the proposed energy router's performance under diverse operating conditions. Experimental results confirm the series module's functionality in both AC and DC grids as an effective solution for controlling extended grids, including power sharing, voltage, and power quality.

eess.SY↗

arXiv2025

AI-Driven Phase-Shifted Carrier Optimization for Cascaded Bridge Converters, Modular Multilevel Converters, and Reconfigurable Batteries

Phase-shifted carrier pulse-width modulation (PSC-PWM) is a widely adopted scheduling algorithm in cascaded bridge converters, modular multilevel converters, and reconfigurable batteries. However, non-uniformed pulse widths for the modules with fixed phase shift angles lead to significant ripple current and output-voltage distortion. Voltage uniformity instead would require optimization of the phase shifts of the individual carriers. However, the computational burden for such optimization is beyond the capabilities of any simple embedded controller. This paper proposes a neural network that emulates the behavior of an instantaneous optimizer with significantly reduced computational burden. The proposed method has the advantages of stable performance in predicting the optimum phase-shift angles under balanced battery modules with non-identical modulation indices without requiring extensive lookup tables, slow numerical optimization, or complex controller tuning. With only one (re)training session for any specified number of modules, the proposed method is readily adaptable to different system sizes. Furthermore, the proposed framework also includes a simple scaling strategy that allows a neural network trained for fewer modules to be reused for larger systems by grouping modules and adjusting their phase shifts. The scaling strategy eliminates the need for retraining. Large-scale assessment, simulations, and experiments demonstrate that, on average, the proposed approach can reduce the current ripple and the weighted total harmonic distortion by up to 50 % in real time and is 100 to 500 thousand times faster than a conventional optimizer (e.g., genetic algorithms), making it the only solution for an online application.

eess.SY↗

arXiv2025

High-Performance Rotor Cooling with Ducted Liquid in Completely Cold-Formed Modular Motor Shaft

This paper suggests a novel rotor-cooling shaft concept for high-performance electric motors that increases the effectiveness of cooling and is yet simple and cost-effective to manufacture. We investigate the thermal performance of four shaft geometries for rotor cooling in automotive applications. The proposed tooth-guided liquid-cooling shaft design aims to solve the high churning loss of conventional cooled rotor shafts due to internal vortex formation and their still limited heat transfer. Therefore, we optimize heat transfer efficiency and pressure management by incorporating cold-formed internal channels that restrict vortex formation beyond a degree that improves heat transfer. We evaluated key performance metrics, including heat transfer rate, outlet temperature, pressure drop, and velocity profiles, under varying rotational speeds, inlet flow rates, and coolant temperatures. Computational fluid analysis demonstrates that the tooth-guided design outperforms conventional hollow shafts and achieves up to 110% higher cooling efficiency at low rotational speeds, while it maintains comparable pressure levels. These findings provide practical insight into geometry-driven thermal optimization and offer a path toward improving the performance and durability of electric motors.

eess.SY↗

arXiv2025

HRTFformer: A Spatially-Aware Transformer for Individual HRTF Upsampling in Immersive Audio Rendering

Individual Head-Related Transfer Functions (HRTFs) are starting to be introduced in many commercial immersive audio applications and are crucial for realistic spatial audio rendering. However, one of the main hesitations regarding their introduction is that creating individual HRTFs is impractical at scale due to the complexities of the HRTF measurement process. To mitigate this drawback, HRTF spatial upsampling has been proposed with the aim of reducing the measurements required. While prior work has seen success with different machine learning (ML) approaches, these models often struggle with long-range preservation of local spatial variation patterns across neighbouring source directions and generalization at high upsampling factors. In this paper, we propose a novel transformer-based architecture for HRTF upsampling, leveraging the attention mechanism to better capture spatial correlations across the HRTF sphere. Working in the spherical harmonic (SH) domain, our model learns to reconstruct high-resolution HRTFs from sparse input measurements with significantly improved accuracy. To enhance spatial coherence, we introduce a neighbour dissimilarity loss that promotes magnitude smoothness, yielding more realistic upsampling. We evaluate our method using both perceptual localization models and objective spectral distortion metrics. Experiments show that our model outperforms existing methods across several evaluation metrics in generating realistic, high-fidelity HRTFs.

cs.SD↗

arXiv2025

Statistical-Spatial Model for Motor Potentials Evoked Through Transcranial Magnetic Stimulation for the Development of Closed-Loop Procedures

The primary motor cortex appears to be in the center of transcranial magnetic stimulation (TMS). It is one of few locations that provide directly observable responses, and its physiology serves as model or reference for almost all other TMS targets, e.g., through the motor threshold and spatial targeting relative to its position. It furthermore sets the safety limits for the entire brain. Its easily detectable responses have led to closed-loop methods for a range of aspects, e.g., for automated thresholding, amplitude tracking, and targeting. The high variability of brain stimulation methods would substantially benefit from fast unbiased closed-loop methods. However, the development of more potent methods would early on in the design phase require proper models that allowed tuning and testing with sufficient without a high number of experiments, which are time-consuming and expensive or even impossible at the needed scale. On the one hand, theoretical researchers without access to experiments miss realistic spatial response models of brain stimulation to develop better methods. On the other hand, subjects should potentially not be exposed to early closed-loop-methods without sufficient prior testing as not yet well tuned feed-back as needed for closed-loop operation is known to erratic behavior. To bridge this gap, we developed a digital-twin-style population model that generates motor evoked potentials in response to virtual stimuli and includes statistical information on spatial (coil position and orientation) as well as recruitment in the population to represent inter- and intra-individual variability. The model allows users to simulate different subjects and millions of runs for software-in-the loop testing. The model includes all code to stimulate further development.

q-bio.NC↗

arXiv2025

Enhanced Fault Ride-Through Grid Forming with Transient Synchronisation Stability and Current Saturation

During grid faults, grid-forming converters are typically suggested to switch from a voltage-source to a current-source mode to limit the current and protect the electronics. This transition has the potential for the converter to transiently lose synchronization due to such current saturation. Therefore, this paper proposes an alternative current saturation algorithm to improve transient synchronization stability during mode switching. The algorithm is designed for grid-forming converters to meet low-voltage ride-through (LVRT) requirements and grid-fault standards in addition to transient synchronization stability. Moreover, it limits the converter output current during grid faults with a new control parameter. The presented method introduces converter output virtual fluxes to calculate the current references in the d- and q-axes for the current saturation algorithm to enhance LVRT performance and grid stability. The method exploits the correlation between the converter's virtual fluxes and currents to modify the current saturation levels through real-time converter virtual flux estimation. The adaptive saturation levels ensure precise control and high dynamics during grid faults and facilitate optimal power injection or absorption to support the grid. The proposed current-saturation algorithm is analytically evaluated. Further, hardware-in-the-loop (HIL) experiments validate the effectiveness of the proposed algorithm.

eess.SY↗

arXiv2025

Asymmetric Modular Pulse Synthesizer: A High-Power High-Granularity Electronics Solution for Transcranial Magnetic Stimulation with Practically Any Pulse Shape for Neural Activation Selectivity

Noninvasive brain stimulation can activate neurons in the brain but requires power electronics with exceptionally high power in the mega-volt-ampere and high frequencies in the kilohertz range. Whereas oscillator circuits offered only one or very few pulse shapes, modular power electronics solved a long-standing problem for the first time and enabled arbitrary software-based design of the temporal shape of stimuli. However, synthesizing arbitrary stimuli with a high output quality requires a large number of modules. Systems with few modules and pulse-width modulation may generate apparently smooth current shapes in the highly inductive coil, but the stimulation effect of the neurons depends on the electric field and the electric field becomes a burst of ultra-brief rectangular pulses. We propose an alternative solution that achieves high-resolution pulse shaping with fewer modules by implementing high-power wide-bandwidth voltage asymmetry. Rather than equal voltage steps, our system strategically assigns different voltages to each module to achieve a near-exponential improvement in resolution. Compared to prior designs, our experimental prototype achieved better output quality, although it uses only half the number of modules.

physics.med-ph↗

arXiv2024

Dynamic Power Management in Modular Reconfigurable Battery Systems with Energy and Power Modules

Integrating power electronics with batteries can offer many advantages, including load sharing and balancing with parallel connectivity. However, parallel batteries with differing voltages and power profiles can cause large circulating currents and uncontrolled energy transfers, risking system instability. To overcome these challenges, we propose a novel modular reconfigurable topology for AC batteries with coupled inductors between adjacent submodules. This approach dynamically manages energy distribution between energy and power modules. Under normal conditions, energy modules operate in series or parallel configurations to deliver stable voltage and current, ensuring efficient power delivery. During heavy loads, such as during electric-vehicle acceleration, power modules seamlessly meet additional power demands beyond the energy modules' capacity. The use of mutual inductors reduces inductance along the load path for faster response times while providing sufficient inductance to regulate circulating currents and enable efficient energy transfer among modules. The system's AC load further complicates control, necessitating a simple yet effective feedforward-feedback control strategy to maintain satisfactory performance.

eess.SY↗

arXiv2024

Three mechanistically different variability and noise sources in the trial-to-trial fluctuations of responses to brain stimulation

Motor-evoked potentials (MEPs) are among the few directly observable responses to external brain stimulation and serve a variety of applications, often in the form of input-output (IO) curves. Previous statistical models with two variability sources inherently consider the small MEPs at the low-side plateau as part of the neural recruitment properties. However, recent studies demonstrated that small MEP responses under resting conditions are contaminated and over-shadowed by background noise of mostly technical quality, e.g., caused by the amplifier, and suggested that the neural recruitment curve should continue below this noise level. This work intends to separate physiological variability from background noise and improve the description of recruitment behaviour. We developed a triple-variability-source model around a logarithmic logistic function without a lower plateau and incorporated an additional source for background noise. Compared to models with two or fewer variability sources, our approach better described IO characteristics, evidenced by lower Bayesian Information Criterion scores across all subjects and pulse shapes. The model independently extracted hidden variability information across the stimulated neural system and isolated it from background noise, which led to an accurate estimation of the IO curve parameters. This new model offers a robust tool to analyse brain stimulation IO curves in clinical and experimental neuroscience and reduces the risk of spurious results from inappropriate statistical methods. The presented model together with the corresponding calibration method provides a more accurate representation of MEP responses and variability sources, advances our understanding of cortical excitability, and may improve the assessment of neuromodulation effects.

q-bio.NC↗

arXiv2024

"You still have to study" -- On the Security of LLM generated code

We witness an increasing usage of AI-assistants even for routine (classroom) programming tasks. However, the code generated on basis of a so called "prompt" by the programmer does not always meet accepted security standards. On the one hand, this may be due to lack of best-practice examples in the training data. On the other hand, the actual quality of the programmers prompt appears to influence whether generated code contains weaknesses or not. In this paper we analyse 4 major LLMs with respect to the security of generated code. We do this on basis of a case study for the Python and Javascript language, using the MITRE CWE catalogue as the guiding security definition. Our results show that using different prompting techniques, some LLMs initially generate 65% code which is deemed insecure by a trained security engineer. On the other hand almost all analysed LLMs will eventually generate code being close to 100% secure with increasing manual guidance of a skilled engineer.

cs.SE↗

arXiv2023

Formalism for Anatomy-Independent Projection and Optimization of Transcranial Magnetic Stimulation Coils

Transcranial magnetic stimulation (TMS) is a popular method for the noninvasive stimulation of neurons in the brain. It has become a standard instrument in experimental brain research and is approved for a range of diagnostic and therapeutic applications. Various applications have been established or approved for specific coil designs with their corresponding spatial electric field distributions. However, the specific coil implementation may no longer be appropriate from the perspective of material and manufacturing opportunities or considering the latest understanding of how to achieve induced electric fields in the head most efficiently. Furthermore, in some cases, field measurements of coils with unknown winding or a user-defined field are available and require an actual implementation. Similar applications exist for magnetic resonance imaging coils. This work aims at introducing a formalism that is completely free from heuristics, iterative optimization, and ad-hoc or manual steps to form practical stimulation coils with a winding consisting of individual turns to either equivalently match an existing coil or produce a given field. The target coil can reside on practically any sufficiently large or closed surface adjacent to or around the head. The method derives an equivalent field through vector projection. In contrast to other coil design or optimization approaches recently presented, the procedure is an explicit forward Hilbert-space vector projection or basis change. For demonstration, we map a commercial figure-of-eight coil as one of the most widely used devices and a more intricate coil recently approved clinically for addiction treatment (H4) onto a bent surface close to the head for highest efficiency and lowest field energy. The resulting projections are within 4% of the target field and reduce the necessary pulse energy by more than 40%.

physics.med-ph↗

arXiv2023

Design and Implementation of DC-to-5~MHz Wide-Bandwidth High-Power High-Fidelity Converter

Advances in power electronics have made it possible to achieve high power levels, e.g., reaching GW in grids, or alternatively high output bandwidths, e.g., beyond MHz in communication. Achieving both simultaneously, however, remains challenging. Various applications, ranging from efficient multichannel wireless power transfer to cutting-edge medical and neuroscience applications, are demanding both high power and wide bandwidth. Conventional inverters can achieve high power and high quality at grid or specific frequency ranges but lose their fidelity when reaching higher output frequencies. Resonant circuits can promise a high output frequency but only a narrow bandwidth. We overcome the hardware challenges by combining gallium-nitride (GaN) transistors with modular cascaded double-H bridge circuits and control that can manage typical timing and balancing issues. We developed a lightweight embedded control solution that includes an improved look-up-table digital synthesizer and a novel adaptive-bias-elimination nearest-level modulation. This solution effectively solves the conflict between a high power level and high output bandwidth and can--in contrast to previous approaches--in principle be scaled in both dimensions. Our prototype exhibits a frequency range from DC to 5 MHz with <18% total voltage distortion across the entire frequency spectrum, while achieving a power level of >5 kW. We conducted tests by sweeping the output frequency and two channel-mixing trials, which included a practical magnetogenetics-oriented stimulation pulse and an entertaining trial to reproduce the famous Arecibo message with the current spectrum.

eess.SP↗

arXiv2023

Topology, Analysis, and Modulation Strategy of a Fully Controlled Modular Battery Pack with Interconnected Output Ports

Modular battery integrated converters or so called dynamically reconfigurable battery packs are expanding into emerging applications, including electromobility and grid storage. Although they offer many degrees of freedom, the state of the art focuses on single output systems and mostly neglects potential of such systems in generating multiple controlled outputs. This paper investigates the use of the extra degrees of freedom that a reconfigurable battery offers to generate multiple outputs for the adjustable supply of various independent loads. The technology offers the potential for functional integration, exploiting already existing transistors, and for a reduction of separate dedicated power conversion stages. Interleaved output ports, where some modules are shared among different outputs, can further reduce the necessary power electronics and cost. However, using conventional modulation techniques in particular phase-shifted carrier modulation can be challenging with shared modules among multiple ports, and different control objectives can adversely impact the overall performance. Therefore, this paper proposes a strategy to decouple the control of multiple ports to enable further simplification of required power electronics. The proposed system does not require any additional active switches for the isolated port and can operate with a wide range of output voltages. Simulation and experiments verify the developed analysis.

eess.SY↗

arXiv2022

A Simplified Model for the Battery Ageing Potential Under Highly Rippled Load for Battery Management and Active Degradation Control

Whereas in typical standardized tests batteries are almost exclusively loaded with constant current or relatively slowly changing cycles, real applications involve rapid load ripple, which do not contribute to the net energy. The trend to reduced filter capacitors and even dynamically reconfigurable batteries further increases the ripple. The influence of rippled load on lithium batteries is therefore receiving increased attention. According to recent studies, accelerated ageing strongly depends on the frequency of the ripple. We use electrochemical models to derive a highly simplified regression model that catches the asymptotic behavior and allows parameter identification and calibration to specific cells. The model allows quantitative monitoring of the additional ageing due to ripple current in battery management systems. Furthermore it enables active control of the ageing potential by influencing the frequency content in modern battery systems, such as reconfigurable batteries.

eess.SY↗