Resource
2026 EN
Yuan Gao · Mattia Piccinini · Yuchen Zhang
+12 more
Ensuring the safety of autonomous vehicles in real-world environments requires handling a wide spectrum of diverse and rare driving scenarios. Scenario-based testing addresses this need by offering a scalable and controlled approach to develop and validate autonomous driving systems. However, traditional scenario generation methods relying on rule-based logic, knowledge-driven models, or data-driven synthesis often yield limited diversity and unrealistic cases. With the emergence of foundation models, which represent a new generation of pre-trained, general-purpose Artificial Intelligence (AI) models, developers can process heterogeneous inputs (e.g., natural language, sensor data, maps, and control actions), enabling the synthesis, interpretation, analysis of complex driving scenarios. In this paper, we review the use of foundation models for scenario generation and scenario analysis in autonomous driving. Our survey presents a unified taxonomy that includes large language models, vision language models, multimodal large language models, diffusion models, and world models for the generation and analysis of autonomous driving scenarios, outlining their fundamental principles, applications, and corresponding evaluation metrics. In addition, we review the methodologies, open-source datasets, simulation platforms, and benchmark challenges. Finally, the survey concludes by highlighting the open challenges, research questions and promising future directions in applying foundation models to scenario generation and analysis in autonomous driving. All reviewed papers are listed in a continuously maintained repository, which is publicly available and updated with new research: GitHub.com/TUM-AVS/FM-for-Scenario-Generation-Analysis.
Resource
2026 EN
Can Wang · Yanming Zhang · Wei Liu
+4 more
This article presents a physics-aware convolutional long short-term memory (PC-LSTM) network for efficient and accurate modeling of mutual impedance matrices in dipole antenna arrays. By reinterpreting Green’s function through a physics-aware neural network (PANN) and embedding it into an adaptive loss function, the proposed machine learning-based approach achieves enhanced physical interpretability in mutual coupling modeling. Also, an attention mechanism is carefully designed to calibrate complex-valued features by fusing the real and imaginary parts of Green’s function matrix. These fused representations are subsequently fed into a convolutional long short-term memory network incorporating a physics-aware convolution kernel, and the impedance matrix of the linear antenna array can be finally derived. Validation against five benchmarks underscores the efficacy of the proposed approach, demonstrating accurate impedance extraction with up to a $7 $ speedup compared to CST Microwave Studio, making it a fast alternative to full-wave simulations for mutual coupling characterization.
Resource
2026 EN
Yilin Gao · Hongbin Zhu · Ning Ma
+4 more
A compact dual-band balanced MIMO antenna is presented in this article for Wi-Fi 6/6E applications in metal-housed tablet computers. To address the challenges of small clearance and metal housing, a systematic evolution of the proposed design, derived from a slot antenna pair with zero edge-to-edge spacing, highlights four key contributions. Firstly, two balanced slot modes, are deliberately excited to support consistent coverage and maintain robust performance against environmental detuning. Secondly, coupled-loaded C-shaped rings (CSRs) are first employed to simultaneously achieve a 59.5% size reduction and generate a transmission zero (TZ) via Fano-type resonance in the 6-GHz band. Thirdly, an additional TZ is realized through modal cancellation, using a chip capacitor, to enhance isolation in the 5-GHz band. Consequently, dual-band coverage and decoupling across Wi-Fi 6/6E bands (5.15–5.83 GHz and 5.925–7.125 GHz) are achieved within a compact antenna footprint of 0.21×0.07 λ L 2 . Finally, an 8×8 MIMO antenna system prototype is fabricated and experimentally verified in a metal-housed tablet platform. Measured results confirm a –6 dB impedance bandwidth (IBW) from 4.3 GHz to 7.5 GHz. Within Wi-Fi 6/6E bands, measured total efficiency from 53.0% to 73.1% and isolation over 12.9 dB are obtained, with acceptable diversity performance.
Resource
2026 EN
Steven L. Kosier · Arijit Sengupta · Sajal Islam
+8 more
Ideal one-dimensional expressions for Single Event Burnout (SEB) Voltage and Single Event Leakage Current (SELC) Voltage in vertical SiC power devices are derived. The derivation employs the critical energy storage and release model to show that the specific on-resistance (R sp ) varies with the SEB and SELC voltages to the 5 th power. Experimental data from MOSFETs and JBS diodes with voltage ratings of 1.2 kV to 10 kV are used to validate and illustrate the ideal limit. A physical understanding of the need for voltage derating is provided.
Resource
2026 EN
Yu Deng · Yiyang Xu · Linglong Qian
+10 more
Cardiac Magnetic Resonance (CMR) imaging is widely used to personalize heart models for cardiac digital twin analysis because of its ability to visualize soft tissues and capture dynamic functions. However, CMR images have an anisotropic nature, characterized by large inter-slice distances and misalignments from cardiac motion. These limitations result in data loss and measurement inaccuracies, hindering the capture of detailed anatomical structures. In this work, we introduce MorphiNet, a novel network that reproduces heart anatomy learned from high-resolution Computed Tomography (CT) images, unpaired with CMR images. MorphiNet encodes the anatomical structure as gradient fields, deforming template meshes into patient-specific geometries. A multilayer graph subdivision network refines these geometries while maintaining dense point correspondence, suitable for downstream computational analysis. MorphiNet achieved the strongest overall trade-off in bi-ventricular myocardium reconstruction on CMR patients with tetralogy of Fallot, with 0.3 higher Dice score and 2.6 lower Hausdorff distance compared to the best existing template-based methods, while achieving comparable geometric accuracy to neural implicit function methods on CT data at 50× faster inference. Cross-dataset validation on the Automated Cardiac Diagnosis Challenge confirmed robust generalization, achieving a 0.7 Dice score with 30% improvement over previous template-based approaches. We validate our anatomical learning approach through the successful restoration of missing cardiac structures and demonstrate significant improvement over standard Loop subdivision. Motion tracking experiments further confirm MorphiNet’s capability for cardiac function analysis, including ejection-fraction estimates that correctly identify myocardial dysfunction in tetralogy of Fallot patients. Code and checkpoints are available at https://github.com/MalikTeng/MorphiNetV2.
Resource
2026 EN
Yun Zhao · Qinlin Gu · Georgios I. Angelis
+3 more
Dynamic total body positron emission tomography (TB-PET) makes it feasible to measure the kinetics of the tracer in all organs of the body simultaneously which may lead to important applications in multi-organ disease and systems physiology. Since whole-body kinetics are highly heterogeneous with variable signal-to-noise ratios, parametric images should ideally comprise not only point estimates but also measures of posterior statistical uncertainty. However, standard Bayesian techniques, such as Markov chain Monte Carlo (MCMC), are computationally prohibitive at the total body scale. We introduce a generative consistency model (CM) that generates samples from the posterior distributions of the kinetic model parameters given measured time-activity curves and arterial input function. CM is able to collapse the hundreds of iterations required by standard diffusion models into just 3 denoising steps. The CM was evaluated using physiologically realistic simulations and an application to a subject’s dynamic [ 18 F]FDG TB-PET dataset analyzed with a standard single-input two-tissue compartment model. When trained on 500,000 physiologically realistic two-tissue compartment model simulations, the CM produces similar accuracy to MCMC (median absolute percent error < 5%; median K-L divergence < 0.5) but is more than five orders of magnitude faster. CM produces more reliable K i images than the Patlak method by avoiding the assumption of irreversibility, while also offering valuable information on statistical uncertainty of parameter estimates and the underlying model. The proposed framework removes the computational barrier to routine, fully Bayesian parametric imaging in TB-PET and is readily extensible to other tracers and compartment models.
Resource
2026 EN
Matteo Pantano · Teresa Wolf · Philipp Mathea
+7 more
Use of endovascular therapy continues to expand due to its effectiveness and minimally invasive characteristics, prompting growing interest in teleoperated robotic systems that can extend access in rural settings and improve physicians well-being. These robotic platforms support physicians in the precise manipulation of endovascular devices, yet the importance for providing haptic feedback within such systems remains a topic of debate. Although several recent works outline techniques for delivering force cues to physicians, comparatively little attention has been given to how these forces are sensed on the robotic side. To clarify this aspect, this study reports a systematic literature review conducted under PRISMA guidelines, supplemented with a bias assessment to ensure cross-database consistency. From an initial pool of 641 publications, eight principal categories of endovascular forces were identified, with reported values spanning $0.17~}$ to $2.61~}$ and occasional peaks reaching $25~}$ . Most measurements were obtained through sensors integrated either on the robotic drive mechanisms or directly onto endovascular devices. In addition, the review highlights emerging directions, including human limits on force sensing, hybrid sensing–estimation methods and the need for force characterization in therapeutic devices. Addressing these aspects will be essential to support future force feedback strategies and improve force assessments in both robot-assisted and manual endovascular procedures.
Resource
2026 EN
Michael D. Adkins · Tyler J. Gourley · Tyler S. Davis
+7 more
Hand dexterity assessments play a crucial role in informing the rehabilitative care of individuals with upper-limb hemiparesis. However, current assessments often struggle to evaluate the hand's ability to precisely control grip force, a skill vital for daily activities like handling fragile objects. Here we describe the design of the Electronic Grip Gauge (EGG), an adjustable-weight, instrumented “fragile” object that measures grip force, load force, acceleration, orientation, and relative position. Embedded sensors enable automatic segmentation and analysis of EGG transfers in various modes. In “Non-Fragile” mode, there is no break threshold; the EGG serves as an automated variant of the Box-and-Blocks test. In “Fragile” mode, the EGG simulates fragility by playing a “break” noise if grip force exceeds a set threshold, requiring grip control to prevent breaks. In “Fragile-Feedback” mode, audio-visual feedback is provided proportional to applied grip force to supplement potentially impaired tactile feedback. Demonstrating functionality, we evaluated sensorimotor differences between 26 hemiparetic and 26 age-matched healthy participants. In “Fragile” mode, paretic hands were significantly slower, applied excessive force, and broke the EGG more frequently than contralateral and healthy control hands. In “Fragile-Feedback” mode, a subset of paretic hands improved, transferring the EGG faster and/or with less force. This work demonstrates the EGG's utility in automatically quantifying sensorimotor deficits and that, for a subset of hemiparetic patients, audiovisual feedback could potentially coach and rehabilitate hand function. Collectively, this work showcases the EGG's potential as both an assessment and rehabilitation device for grip force control – a critical skill in hand therapy.
Resource
2026 EN
Sara C. Doan · Steven S. Brooks
Background: This article presents a systematic review of data visualization scholarship in technical and professional communication (TPC) from 2013 to 2023. For interdisciplinary fields, such as TPC, where scholars draw on a wide range of epistemologies and methodologies, identifying patterns and trends prevents scholars from unknowingly talking past each other. Literature review: Our study was grounded in categorization scholarship in technical communication and contemporary issues the field faces. Research questions: 1. What kinds of articles are being published on data visualizations in TPC? 2. What case studies or examples do TPC scholars use to study data visualizations and from which disciplines or domains do they draw? 3. How can we categorize the types of data visualization written about in TPC journals? 4. What can these findings tell us about the current state of data visualizations in TPC journals and where further attention might be needed? Methodology: We conducted a content analysis of 94 articles across seven leading TPC journals. We developed a code book to identify common subject themes across scholarship looking at various types of data visualization types and topics. Results: Our results explore the field's growing emphasis on socially engaged themes, such as health, risk, and environmental justice. Conclusion: We call TPC to turn attention to accessibility and pedagogy.
Resource
2026 EN
Jorge Roser · Hong Phuc Vo · Rebecca Kantorek
+2 more
Inter-crystal scattering (ICS) in Positron Emission Tomography (PET) is commonly regarded as a degradation effect that might compromise the image spatial resolution. In parallel, the inclusion of ICS events has also been recognized as a potential approach to increase PET sensitivity, which could be especially beneficial in scenarios where the latter is a limiting factor, such as very small animal imaging. Several methods for the recovery of ICS events have been proposed, many of which aim to locate the first interaction, i.e., the Compton scattering site, usually limited by their success rate, computational burden or data and training dependency. Conversely, this work proposes a physics-based model for ICS events, leading to analytical expressions of the sensitivity image and the system matrix (required by statistical reconstruction algorithms), without the need to identify the original line of response. After validating the model, the work shows how ICS events can be integrated into a joint image reconstruction algorithm (based on list-mode MLEM) together with conventional PET events, for which dedicated analytical models are also developed. To assess the performance of the proposed approach, Monte-Carlo simulated and experimental data of an image quality phantom were obtained with the MERMAID small-fish PET scanner prototype. Both simulation and experimental results indicate that, while slightly decreasing the recovery coefficient values, the inclusion of ICS clearly reduces statistical noise and improves uniformity.