Computing matrix gradient has become a key aspect in modern signal processing/machine learning, with the recent use of matrix neural networks requiring matrix backpropagation. In this field, two main methods exist to calculate the gradient of matrix functions for symmetric positive definite (SPD) matrices, namely, the Daleckiǐ-Kreǐn/Bhatia formula and the Ionescu method. However, there appear to be a few errors. This brief aims to demonstrate each of these formulas in a self-contained and unified framework, to prove theoretically their equivalence, and to clarify inaccurate results of the literature. A numerical comparison of both methods is also provided in terms of computational speed and numerical stability to show the superiority of the Daleckiǐ-Kreǐn/Bhatia approach. We also extend the matrix gradient to the general case of diagonalizable matrices. Convincing results with the two backpropagation methods are shown on the EEG-based BCI competition dataset with the implementation of an SPDNet, yielding around 80% accuracy for one subject. Daleckiǐ-Kreǐn/Bhatia formula achieves an 8% time gain during training and handles degenerate cases.

With the increasing demand for dynamic behaviors in automotive use cases, Software Defined Vehicles (SDVs) have emerged as a promising solution by bringing dynamic onboard service management capabilities. While users may request a wide range of services during vehicle operation, background tasks such as cooperative Vehicle-to-Everything (V2X) services can activate on-the-fly in response to real-time road conditions. In this dynamic environment, the efficient allocation of onboard resources becomes a complex challenge, in order to meet mixed-criticality onboard Quality-of-Service (QoS) network requirements while ensuring an optimal user experience (UX). Additionally, the ever-evolving real-time network connectivity and computational availability conditions further complicate the process. In this context, first we introduce the concept of Automotive eXperience Integrity Level (AXIL) which expresses a runtime priority for non-safety-critical applications to enable dynamic orchestration. Second, we present a dynamic resource-based onboard service orchestration algorithm that considers real-time in-vehicle and V2X network health, along with onboard resource constraints, to select degraded modes for onboard applications and maximize user experience. Third, we show by simulation that our algorithm produces near-optimal solutions while significantly reducing execution time: it satisfies approximately 90% of the onboard UX compared to the optimal while reducing the execution time of one to two orders of magnitude compared to an exact solver. With this approach, we aim to enable efficient onboard execution for a UX-focused service orchestration.

The increasing connectivity and autonomy of vehicles has led to a growing need for dynamic and real-time adjustments to software and network configurations. Software Defined Vehicles (SDV) have emerged as a potential solution to adapt to changing user needs with continuous updates and onboard reconfigurations to offer infotainment, connected, and background services such as cooperative driving. However, network configuration management in SDVs remains a significant challenge, particularly in the context of shared Ethernet-based in-vehicle networks. Traditional worst-case static configuration methods cannot efficiently allocate network resources while ensuring Quality of Service (QoS) guarantees for each network flow within the physical topology capabilities. In this work, we propose a configuration generation methodology that addresses these limitations by dynamically switching between pre-computed offboard configurations downloaded to the vehicle. Simulation results are presented and future work is discussed.

Motor Brain-Computer Interfaces (BCIs) are systems that allow severely motor-impaired patients to use their brain activity to interact with their environment. Electrocorticography (ECoG) arrays may be profitably used to develop safe and chronic motor BCI systems. BCI signal processing pipelines generally include neuronal signal pre-processing, feature extraction and classification/regression. The article presents a comparative study addressing the problem of neural feature classification in asynchronous multi-limb ECoG-driven BCIs. Several conventional classifiers often reported in the BCI literature were coupled with two preprocessing techniques and with a conventional feature extraction approach. They were compared to artificial neural network (ANN) end-to-end classifiers which mimic conventional BCI signal processing pipelines. Different initializations of ANNs were particularly studied. The comparison study was carried out using publicly available datasets (BCI competition IV).

One of the main ideas in the area of intelligent transport systems is to use all possible information coming from vehicles and infrastructure, in order to make the system “smarter” and avoid potentially dangerous situations-collisions, accidents, bottleneck, etc. However, data are sometimes unreliable due to source and communication network quality, leading vehicles or even the whole system to wrong decisions. We present a generic method for detecting dangerous events on the road. To support unreliable data sources, it uses distributed data fusion. Moreover, to deal with network failures, it relies on a self-stabilizing generic distributed algorithm. Our method mixes measurements obtained from vehicle onboard sensors, as well as wireless sensors placed close to the road and connected to road side units. Each vehicle computes how confident it is about a potential dangerous event using both local and remote data. To evaluate our approach, we implemented it using a specific hardware and software platform. Moreover, we instantiated a simple, yet efficient application to detect icy roads, based on temperature measurements. Thanks to both in-lab and actual on-the-road experiments, we demonstrate the possibility to deduce proper results from unreliable data and, consequently, the correctness and usefulness of our approach.

This paper presents a link between the well known Common Spatial Pattern (CSP) algorithm and Riemannian geometry in the context of Brain Computer Interface (BCI). It will be shown that CSP spatial filtering and Log variance features extraction can be resumed as a computation of a Riemann distance in the space of covariances matrices. This fact yields to highlight several approximations with respect to the space topology. According to these conclusions, we propose an improvement of classical CSP method.

Erosion in active mountain belts is one of the key questions in geomorphology. Remote sensing can be a powerful tool to predict erosion on a quantitative scale and helps to understand the near-surface processes involved. One of the major limits in these kinds of environments is the lack of ground truth information and its inaccessibility. Here we present an integrated approach in order to validate remote sensing applications with on-site information in small catchments. Image classification (MLC) on SPOT-5 data is applied to map land-use and relate it spatially with typical erosion rates. Precipitation measurements from the Tropical Rain Measuring Mission (TRMM) are compared with real precipitations from a rain gauge network. Denudation rates derived from land-use classifications (soil erosion) are related to absolute denudation rates calculated from suspended sediment concentrations. Our results outline the intricacy of competing processes in active mountain environments and that it is possible to measure spatial erosion from remote sensing information. Nevertheless, the spatial resolution, especially of TRMM data is limiting the small-scale erosion prediction.

This paper presents the development and hardware experimentations for the attitude control of a UAV. The AURYON system consists of a particular contrarotating coaxial rotors aircraft, an embedded and a ground control systems, a distributed real-time 3D simulator based on The Mathworks xPC-Target toolbox and Laminar Research X-Plane flight simulator. The controller developed for the attitude and altitude stabilization is based on a two-rule fuzzy controller. A parameter-tuning procedure, using a real-time 3D simulator is proposed. The performances of the UAV mechatronic system are illustrated trough real demonstrator experimentations.

Dynamic computed tomography (CT) imaging aims at reconstructing image sequences where the dynamic nature of the living human body is of primary interest. The main applications concerned are image-guided interventional procedures, functional studies and cardiac imaging. The introduction of ultra-fast rotating gantries along with multi-row detectors and in near future area detectors allows huge progress toward the imaging of moving organs with low-contrast resolution. This paper gives an overview of the different concepts used in dynamic CT. A new reconstruction algorithm based on a voxel-specific dynamic evolution compensation is also presented. It provides four-dimensional image sequences with accurate spatio-temporal information, where each frame is reconstructed using a long-scan acquisition mode on several half-turns. In the same time, this technique permits to reduce the dose delivered per rotation while keeping the same signal to noise ratio for every frame using an adaptive motion-compensated temporal averaging. Results are illustrated on simulated data.

In this paper, we propose a new wavelet-based reconstruction method suited to three-dimensional (3-D) cone-beam (CB) tomography. It is derived from the Feldkamp algorithm and is valid for the same geometrical conditions. The demonstration is done in the framework of nonseparable wavelets and requires ideally radial wavelets. The proposed inversion formula yields to a filtered backprojection algorithm but the filtering step is implemented using quincunx wavelet filters. The proposed algorithm reconstructs slice by slice both the wavelet and approximation coefficients of the 3-D image directly from the CB projection data. The validity of this multiresolution approach is demonstrated on simulations from both mathematical phantoms and 3-D rotational angiography clinical data. The same quality is achieved compared with the standard Feldkamp algorithm, but in addition, the multiresolution decomposition allows to apply directly image processing techniques in the wavelet domain during the inversion process. As an example, a fast low-resolution reconstruction of the 3-D arterial vessels with the progressive addition of details in a region of interest is demonstrated. Other promising applications are the improvement of image quality by denoising techniques and also the reduction of computing time using the space localization of wavelets.

In this paper, a new multiresolution reconstruction approach for fan-beam tomography is established. The theoretical development assumes radial wavelets. An approximate reconstruction formula based on a near-radial quincunx multiresolution scheme is proposed. This multiresolution algorithm allows to compute both the quincunx approximation and detail coefficients of an image from its fan-beam projections. Simulations on mathematical phantoms show that wavelet decomposition is acceptable for small beam angles but deteriorates at high angles. The main applications of the method are denoising and wavelet-based image analysis.

In this paper, the use of nonseparable wavelets for tomographic reconstruction is investigated. Local tomography is also presented. The algorithm computes both the quincunx approximation and detail coefficients of a function from its projections. Simulation results showed that nonseparable wavelets provide a reconstruction improvement versus separable wavelets.