Autonomous vehicles rely on HD maps for their operation, but offline HD maps eventually become outdated. For this reason, online HD map construction methods use live sensor data to infer map information instead. Research on real map changes shows that oftentimes entire parts of an HD map remain unchanged and can be used as a prior. We therefore introduce M3TR (Multi-Masking Map Transformer), a generalist approach for HD map completion both with and without offline HD map priors. As a necessary foundation, we address shortcomings in ground truth labels for Argoverse 2 and nuScenes and propose the first comprehensive benchmark for HD map completion. Unlike existing models that specialize in a single kind of map change, which is unrealistic for deployment, our Generalist model handles all kinds of changes, matching the effectiveness of Expert models. With our map masking as augmentation regime, we can even achieve a +1.4 mAP improvement without a prior. Finally, by fully utilizing prior HD map elements and optimizing query designs, M3TR outperforms existing methods by +4.3 mAP while being the first real-world deployable model for offline HD map priors.
Accurate stress monitoring is critical for high-risk professions like firefighting, yet existing wearable solutions face challenges balancing accuracy with practical usability. While electrodermal activity (EDA) offers a non-invasive, single-sensor approach, current automated feature extraction methods fail to capture stress-discriminative patterns effectively. We developed a hybrid stress detection pipeline combining 20 hand-crafted physiological features with 32 deep-learned features from a supervised convolutional autoencoder. Unlike traditional unsupervised approaches optimized solely for signal reconstruction, our architecture employs a dual-head design with weighted classification loss to guide feature learning toward stress discrimination. The system was validated on the WESAD dataset (15 subjects) using rigorous leave-one-subject-out (LOSO) cross-validation, along with comprehensive preprocessing, including cvxEDA decomposition, adaptive artifact detection, and physiological peak validation. Our optimized K-Nearest Neighbors classifier achieved 98.62% accuracy, surpassing the industry-standard PyEDA benchmark (97.0%) by 1.62 percentage points. The model demonstrated 97.58% sensitivity (true positive rate) and 98.92% specificity (true negative rate), with only 2.42% false negatives-critical for safety-critical applications. Ablation studies revealed that unsupervised autoencoder features alone achieved only 55% accuracy, increasing to 89% with supervised learning and 98.62% with the hybrid approach, representing a 43.62-percentage-point improvement. This work demonstrates that combining domain-specific physiological knowledge with label-aware deep learning produces more discriminative features than either approach alone. The resulting system successfully translates complex probabilistic outputs into an interpretable 1-10 stress score, providing a practical foundation for real-time stress monitoring in wearable devices for first responders.
Biological plausibility is a key concept in neuromorphic computing and spiking neural networks, yet it remains inconsistently defined and difficult to quantify. In this work, we present an open-source framework for the automated assessment of biological plausibility in spiking neuron models. Our method builds on the idea of evaluating a model's ability to replicate canonical neuronal firing patterns observed in biological systems, following the classification proposed by Izhikevich. By encoding these patterns into objective functions and optimizing model parameters accordingly, our framework enables empirical assessment without requiring prior analytical modeling. Treating neuron models as black boxes, it provides a practical and flexible means of characterizing their dynamic capabilities. We demonstrate the effectiveness of the framework on several established models and a previously unexplored custom model. Implemented in Python and compatible with PyTorch and the Norse library, the framework is tailored for machine learning contexts. It is intended as a starting point for systematic research into the relationship between biological plausibility and network-level performance metrics such as accuracy, energy efficiency, robustness, and adaptability.
Autonomous vehicles (AVs) promise safer, cleaner, and more inclusive mobility, yet large-scale adoption is hindered by user acceptance rather than by technical challenges. Prior studies on acceptance and user experience largely rely on surveys, simulators or Wizard-of-Oz setups, often over-representing technologically enthusiastic participants and focusing on drivers instead of passengers. We address this gap with real-world field studies with AVs in real traffic, totaling 144 participants. Using multi-modal sensing, we evaluated EGG, heartbeat, breathing, camera and voice signals for affect inference in combination with vehicle data. Our results show that breathing, camera and voice measurements are reliable and pratical in naturalistic passenger contexts. We further contribute a validated study protocol, a self-assessment app for real-time assessment during human-machine interaction, and a tailored questionnaire to capture participant attitudes towards AVs. By grounding UX evaluation in real-world contexts, this work lays a foundation for user-centered design of autonomous mobility systems and robotics in general. Our work bridges the gap between affective computing and technical implementation of autonomous vehicles.
Recent advances in Connected and Automated Vehicle (CAV) technology have intensified interest in Cooperative Intelligent Transport Systems (C-ITS), in particular, the cooperation between CAVs and smart road infrastructure as a means to manage the complexity of urban traffic. However, most prior work has either been limited to low cooperation levels, e.g., simple status sharing, or has only been tested in simulation. In this paper, we present real-world test results of Managed Automated Driving (MAD). MAD integrates collective perception from infrastructure- and vehicle-mounted sensors with infrastructure-based multi-vehicle trajectory planning to support and/or control automated vehicles in mixed traffic. To the best of our knowledge, this is the first real-world realization of closed-loop infrastructure-based trajectory prescription on public urban roads. Our experiments indicate that, despite perception noise, heterogeneous human-driven traffic, and non-negligible variability in network performance, infrastructure-based prescriptive planning maintains comfortable motion profiles while meeting real-time execution bounds. The results demonstrate the feasibility of infrastructure-supported automated driving on public roads and provide an architectural blueprint for robust C-ITS deployments in complex urban environments.
