Improving performance and optimizing quality and safety for users of transport networks (vehicular, multimedia, cellular networks like 5G, the Internet of Things (IoT), etc.).
Functional systems are generally designed to meet demands requiring the transport of data flows across various environments, considering user profiles and usage contexts. This necessitates the development of tools that optimize service performance while ensuring security, taking into account both endogenous and exogenous parameters of functional and transport systems.
The overarching context of the TEI transverse theme focuses on:
- Applying machine learning techniques to estimate lost data and detect false information in networks (notably vehicular networks).
- Modeling and developing adaptive mechanisms capable of extracting characteristics from the functional systems studied (vehicular, multimedia, etc.). These characteristics represent information derived from a given environment and context and relate to the Quality of Experience (QoE).
- Proposing corrective solutions in response to undesirable events such as unsatisfactory Quality of Service (QoS), negative feedback, or malfunctions in elements of the transport network.
In addition, research activities are conducted in the domain of mixed reality, aiming to develop an immersive and interactive perception system through visual augmentation and natural interaction in a mixed world. Performance optimization ensures relevance, computational efficiency, robustness, and the enhancement of the mixed reality experience by providing a strong sense of presence.
This research focuses on the learning and classification of natural objects in real-world scenes through immersion and interaction paradigms, which are fundamental characteristics for implementing mixed reality systems. Advances in this field can contribute to the real-time creation of reliable and precise maps of environments, enabling the recognition of various visual targets and natural landmarks in diverse contexts such as mobility.
Control and Development of New Power Converter Topologies
In connection with electric modes of transport, another key area of focus involves proposing new topologies of power converters. In recent years, research in power electronics for automotive applications—particularly Electric Vehicles (EVs) and Hybrid Electric Vehicles (HEVs)—has been steadily advancing. Increasingly, new power converter topologies are being developed to improve energy efficiency.
Some members of the TEI transverse theme are conducting research on optimizing energy usage in embedded systems within electric vehicles. Their work centers on power electronics, aiming to innovate in the technological design of power converters used in EVs to make them more reliable, robust, and cost-effective. This, in turn, seeks to enhance consumer adoption of electric vehicles.
Such research involves the development of tools and numerical methods for modeling, simulation, and optimizing control strategies for converters applied in EVs.
Advanced Driver Assistance Systems
Accidentology studies have shown that human errors are responsible for 90% of road accidents. Each year, road accidents claim 1.25 million lives and injure up to 50 million people worldwide. Partially or fully autonomous vehicles are recognized as having the potential to reduce the number and severity of accidents. These technologies have been made possible thanks to advancements in embedded electronics and information processing.
According to the Society of Automotive Engineers (SAE), autonomy levels are classified into six categories, ranging from Level 0 (no automation, where driving is entirely the driver’s responsibility, possibly assisted by visual, auditory, or tactile alerts) to Level 5 (fully automated driving managed by onboard intelligence, also referred to as an autopilot system).
In this context, studies are being conducted by members of the TEI theme on the architecture of Level 5 autonomous vehicles. These studies focus particularly on:
- The vehicle’s embedded electronics,
- Perception algorithms,
- Decision-making and trajectory planning modules, and
- Control/command algorithms.