VeDeCoM Institute

As the automotive industry shifts toward Software-Defined Vehicles, seamless connectivity has become a key enabler of continuous software integration, over-the-air updates, and real-time data exchange. At the center of this transformation is the Telematic Control Unit (TCU), which orchestrates Vehicle-to-Everything (V2X) communications and unifies a wide range of connectivity interfaces. However, maintaining reliable performance under real-time and resource-constrained conditions requires advanced system-level observability. In this paper, we introduce TCU-Observer, a modular and extensible onboard observability framework specifically designed for TCU in connected vehicles. Our architecture supports real-time monitoring, telemetry analysis, and AI-based inference. In addition, we propose CLAE, a lightweight feature engineering model designed for a TCU embedded environment, enabling efficient transformation of raw telemetry into actionable insights. Extensive V2X experimentation validates the feasibility and effectiveness of our approach, demonstrating its potential to enhance the TCU’s connectivity and reliability.

Reliable multi-source fusion is crucial for robust perception in autonomous systems. However, evaluating fusion performance independently of detection errors remains challenging. This work introduces a systematic evaluation framework that injects controlled noise into ground-truth bounding boxes to isolate the fusion process. We then propose Unified Kalman Fusion (UniKF), a late-fusion algorithm based on Kalman filtering to merge Bird's Eye View (BEV) detections while handling synchronization issues. Experiments show that UniKF outperforms baseline methods across various noise levels, achieving up to <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$3 \times$</tex> lower object's positioning and orientation errors and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$2 \times$</tex> lower dimension estimation errors, while maintaining nearperfect precision and recall between 99.5 % and 100 %.

In autonomous driving, recent research has increasingly focused on collaborative perception based on deep learning to overcome the limitations of individual perception systems. Although these methods achieve high accuracy, they rely on high communication bandwidth and require unrestricted access to each agent's object detection model architecture and parameters. These constraints pose challenges real-world autonomous driving scenarios, where communication limitations and the need to safeguard proprietary models hinder practical implementation. To address this issue, we introduce a novel late collaborative framework for 3D multi-source and multi-object fusion, which operates solely on shared 3D bounding box attributes-category, size, position, and orientation-without necessitating direct access to detection models. Our framework establishes a new state-of-the-art in late fusion, achieving up to five times lower position error compared to existing methods. Additionally, it reduces scale error by a factor of 7.5 and orientation error by half, all while maintaining perfect 100% precision and recall when fusing detections from heterogeneous perception systems. These results highlight the effectiveness of our approach in addressing real-world collaborative perception challenges, setting a new benchmark for efficient and scalable multi-agent fusion.

Abstract This paper proposes and tests a methodology to analyse end consumers’ choices in terms of shopping destination for a store selling culture products, comparing a city centre location to a peripheral one. The proposed methodology begins with a stated preferences survey and incorporates a conditional tree classification algorithm to pre-select the predictors (attributes), then used to develop a discrete choice model. To validate the methodology, a real-world case study was carried out, including a survey with over one thousand customer responses. The findings reveal noteworthy insights into customer attitudes toward relocation, distinguishing frequent from non-frequent users and examining factors such as travel distance and visit frequency. These results offer valuable guidance for retailers and policy makers in shaping city logistics scenarios, highlighting the potential transformations in urban freight flows driven by changes in retail land use.

In this paper we address visual based localization in outdoor environments where the appearance changes dra-matically. Such environmental changes result in a substantial transformation of the visual information of the scene, producing a significant impact on the visual based localization performance. Hence, these changes can lead to major difficulties when associating data between the current image and the landmarks in the map. One solution for this problem is to keep adding landmarks to the map in order to cover various environmental conditions. However, this solution leads to a continued growth of the map, which in turn, will result in a costly and resource-intensive localization. In this paper we present a map management approach in which we exploit information related to the suns position to compare resemblance between the traversals in the map and maintain a diverse map that incorporates a minimum amount of data and ensures a reliable localization in different environmental conditions. We evaluated our approach on a dataset that incorporates more than 100 sequences with different environmental conditions and we compared the obtained results with a state of the art approach.

The current charging technology for electric vehicles consists of plugging the cable from the AC utility to charge the batteries. This requires heavy gauge cables to connect to electric vehicles, which can be difficult to handle, presents tripping hazards, and is prone to vandalism. In addition to these inconveniences, electric vehicles must be immobilized for hours before being fully charged. Dynamic wireless power transfer has been studied worldwide as a promising technology. It is safe and convenient and allows electric vehicles to charge while moving. To improve the efficiency of a dynamic wireless power transfer system, the magnetic coupling coefficient must be maximized between the primary pad, which is integrated into the road, and the secondary pad installed in the electric vehicle. This article presents a parametric optimization of the ferrite structure used for a 3 kW dynamic wireless power transfer prototype. Different ferrite configurations are compared while studying the effect of the parameter values on their magnetic coupling coefficient. Finally, the proposed structure was validated during the experimental test, and its coupling coefficient was improved by 26% compared to the original structure.

The forthcoming arrival of automated vehicles (AV) on the roads requires the re-evaluation or even adaptation of existing infrastructures as they are currently designed on the basis of human perception. Indeed, advanced driver-assistance systems (ADAS) do not necessarily have the same needs as drivers to detect road markings. One of the main challenges related to AV is the optimisation of the vehicle–infrastructure pair in order to guarantee the safety of all users. In this context, we compared the performance of a vehicle equipped with an ADAS machine-vision system with a dynamic retroreflectometer during the daytime on a road section. Our results questioned the reliability of the literature thresholds of the luminance contrast ratio on a dry road under sunny conditions. Despite the presence of old and worn road markings, the ADAS camera was able to detect the edge lines in more than 90% of the cases. The non-detections were not related to the poor condition of the markings but to the environmental conditions or the complexity of the infrastructure.

Dans le cadre des plateformes multimodales, la définition d’un itinéraire est nécessaire à l’usager pour effectuer un choix parmi les différents modes présents pour arriver à destination. (Brunyé et al., 2010). Dans ce cas, l’information dite « multimodale » est la principale ressource disponible pour effectuer un choix modal. Elle permet de comparer les attributs des différents itinéraires identifiés par l’individu (par exemple, durée, prix) (Tollis et al., 2020). Cet article tente d’identifier trois principaux déterminants associés aux informations sollicitées par l’individu pour adopter un itinéraire « convenable » au sein de ces plateformes partagées. L’objectif sera ainsi de cerner, sur la base de la littérature, les facteurs qui, d’une part sont à l’origine des besoins informationnels de l’usager pour prendre une décision modale, et d’autre part, influencent la nature de ces besoins. Le premier déterminant (1) est inhérent aux critères de mobilité recherchés par l’usager pendant son déplacement. Ce déterminant compte deux dimensions, les préférences de mobilité propres à l’usager qui découlent de ses caractéristiques individuelles, et les caractéristiques du contexte, qui induisent un certain nombre de contraintes à respecter. Le deuxième déterminant (2) est lié aux attributs de l’offre modale des plateformes multimodales en temps réel. Il fait référence aux incertitudes de l’offre, notamment en cas de perturbation de l’itinéraire initialement planifié, mais également à la configuration des plateformes et à leurs propriétés en matière de réseau. Enfin, le troisième déterminant (3) regroupe les filtres de modulation des besoins informationnels. Cette modulation est liée à deux éléments, l’habitude et les stratégies d’économie cognitive, qui permettent à l’usager de limiter les informations nécessaires pour faire un choix d’itinéraire.