洛桑联邦理工学院

Accurate trajectory prediction of vulnerable road users is a cornerstone of safe autonomous driving and intelligent transportation systems. While large-scale pre-training has advanced this field, achieving robust zero-shot generalization remains a critical challenge for real-world deployment, particularly when vehicles encounter unseen environments and heterogeneous sensor configurations (e.g., varying frame rates and observation horizons). In this work, we revisit zero-shot trajectory prediction from the perspective of distribution shifts and distinguish three transfer settings: temporal transfer, scene transfer, and joint scene–temporal transfer. Through systematic experiments, we show that temporal mismatch is a key source of failure in current pre-trained models. By isolating temporal configuration from dataset shift, we demonstrate that explicitly conditioning on temporal metadata provides a simple and highly effective solution. Building on this insight, we propose OmniTraj, a Transformer-based framework pre-trained on large-scale heterogeneous data with explicit temporal-aware design. OmniTraj is designed to handle omni-generalization in trajectory prediction, namely adaptability across temporal configuration and scene shifts. It achieves state-of-the-art zero-shot generalization under joint scene–temporal transfer, reducing prediction error by over 70%. Furthermore, it exhibits exceptional robustness in safety-critical edge cases with severely limited observations and maintains high few-shot data efficiency, paving the way for scalable, dataset-agnostic deployment in real-world autonomous systems. The code is publicly available: https://github.com/vita-epfl/omnitraj.

As the share of renewable energy in power systems increases, the resulting need for additional flexibility can be supported by second-life battery energy storage systems. However, their operation is complicated by the heterogeneity of the constituent battery packs. This paper addresses this challenge by proposing an optimization framework that uses battery packs’ losses as a proxy for battery degradation. The proposed two-stage framework (i) optimizes the participation of multiple second-life battery packs, each connected to its own DC/DC converter, in the Frequency Containment Reserve and day-ahead markets, while enforcing a specific constraint to ensure homogeneous degradation of each pack, and (ii) employs a model predictive control-based convex optimization to track market commitments while equalizing resistive losses among the batteries. The framework is validated in simulation across three scenarios with varying initial conditions, demonstrating effectiveness in balancing performance and degradation among heterogeneous battery packs.

We present an implicit, fully-coupled hydro-mechanical solver for the three-dimensional simulation of fluid-driven rupture propagation along pre-existing discontinuities. The solver simultaneously handles frictional slip and tensile failure along arbitrary intersecting fractures and faults in a linearly elastic and impermeable rock matrix. Spatial discretization combines a displacement discontinuity boundary element method with a Galerkin finite element method for pore-fluid pressure diffusion. Frictional and tensile failure are governed by a poro-elastoplastic interface law incorporating slip-weakening friction, dilatancy, and tensile strength degradation. Block preconditioning of the coupled tangent system ensures robustness across a wide range of fracture behaviors, including friction and tensile hydraulic failure. Solver accuracy is verified – for the first time – against a comprehensive suite of semi-analytical rupture propagation solutions of increasing complexity: fluid-driven frictional ruptures, dilatant ruptures with permeability changes, and penny-shaped hydraulic fractures spanning the viscosity-to-toughness transition. Two multi-fracture examples further demonstrate the solver’s capabilities: injection into intersecting fractures, and a hydraulic fracture intersecting a strike-slip fault. These highlight the ability of the algorithm to capture frictional slip, dilatancy, permeability evolution, and tensile opening within a unified framework, making it well suited for fluid-driven rupture simulation in faulted and fractured rocks.

The pantograph–catenary interface is essential for ensuring uninterrupted and reliable power delivery in electrified rail systems. However, electrical arcing at this interface poses serious risks, including accelerated wear of contact components, degraded system performance, and potential service disruptions. Detecting arcing events at the pantograph–catenary interface is challenging due to their transient nature, noisy operating environment, data scarcity, and the difficulty of distinguishing arcs from other similar transient phenomena. To address these challenges, we propose a novel multimodal framework that combines high-resolution image data with force measurements to more accurately and robustly detect arcing events. First, we construct two arcing detection datasets comprising synchronized visual and force measurements. One dataset is built from data provided by the Swiss Federal Railways (SBB), and the other is derived from publicly available videos of arcing events in different railway systems and synthetic force data that mimic the characteristics observed in the real dataset. Leveraging these datasets, we propose MultiDeepSAD, an extension of the DeepSAD algorithm for multiple modalities with a new loss formulation. Additionally, we introduce tailored pseudo-anomaly generation techniques specific to each data type, such as synthetic arc-like artifacts in images and simulated force irregularities, to augment training data and improve the discriminative ability of the model. Through extensive experiments and ablation studies, we demonstrate that our framework significantly outperforms baseline approaches, exhibiting enhanced sensitivity to real arcing events even under domain shifts and limited availability of real arcing observations. To the best of our knowledge, this is the first method and publicly available dataset that integrates image and force data for pantograph–catenary arcing event detection. The proposed framework offers a practical solution for real-time monitoring of arcing events in pantograph–catenary systems, ultimately contributing to safer and more reliable railway operations. Our source code and dataset are at https://github.com/EPFL-IMOS/Multimodal-Arcing.

Hydrogen produced from biomass and waste via plasma technologies offers a sustainable pathway for clean energy generation while addressing waste management challenges. This review examines recent advancements in plasma pyrolysis and gasification for hydrogen production, comparing thermal (microwave, gliding arc, plasma torches) and non-thermal (dielectric barrier discharge, glow discharge) plasmas. Beyond, summarizing recent developments, this review provides an integrated comparison of thermal and non-thermal plasma technologies for hydrogen-rich syngas production from biomass and waste, linking plasma type, reactor characteristics, reaction pathways, and product quality. It also identifies the main scientific and engineering bottlenecks for industrial deployment, including energy efficiency, reactor scaling, cost structure, and environmental performance. Key findings demonstrate that plasma technologies enhance hydrogen yields (up to 80 wt