Publications

2022

• Settling of localized particle plumes in a quiescent water tank
  
  • Zürner Till
  • Toupoint Clément
  • de Souza David
  • Mezouane Dylan
  • Monchaux Romain
  , 2022. The fall or rise of inertial particles plumes in an initially quiescent fluid is ubiquitous in natural and industrial processes. Disentangling the influence of the numerous parameters of these systems on the plume velocity is thus a major issue. Moreover, understanding the energy transfer between the two phases and the structure of induced flow is another very relevant question. A specific experiment has been designed to tackle these problems. A continuous plume of inertial particles released above the center of a still water tank generates a large-scale recirculation flow. The present experimental study investigates the stationary settling dynamics and induced flow properties using various particle populations with particle-to-fluid density ratios from 3.8 to 14.3, Archimedes numbers 4 to 580 and mass fractions 10^−6 to 10^−3, revealing an unexplored region of the parameter space. The fluid and particle velocities are measured simultaneously by two cameras, utilizing optical filtering and image post-processing. It is found that the settling in the laboratory frame generally follows the terminal velocity predicted using the Schiller-Naumann drag. The particle velocity relative to the surrounding fluid is unaffected by the particle mass fraction and always hindered due to the kinetic energy transferred to the fluid. This transfer of energy is most effective for low Archimedes numbers. All modifications of the settling behavior due to the particle mass fraction are caused by the back-reaction of the induced flow on the particles, no direct particle-particle interactions are observed.
• A hybrid level-set / embedded boundary method applied to solidification-melt problems
  
  • Limare Alexandre
  • Popinet Stéphane
  • Josserand Christophe
  • Xue Zhonghan
  • Ghigo Arthur R.
  Journal of Computational Physics, Elsevier, 2022. In this paper, we introduce a novel way to represent the interface for two-phase flows with phase change. We combine a level-set method with a Cartesian embedded boundary method and take advantage of both. This is part of an effort to obtain a numerical strategy relying on Cartesian grids allowing the simulation of complex boundaries with possible change of topology while retaining a high-order representation of the gradients on the interface and the capability of properly applying boundary conditions on the interface. This leads to a two-fluid conservative second-order numerical method. The ability of the method to correctly solve Stefan problems, onset dendrite growth with and without anisotropy is demonstrated through a variety of test cases. Finally, we take advantage of the two-fluid representation to model a Rayleigh-Bénard instability with a melting boundary. (10.1016/j.jcp.2022.111829)
  DOI : 10.1016/j.jcp.2022.111829
• MUAD: Multiple Uncertainties for Autonomous Driving, a benchmark for multiple uncertainty types and tasks
  
  • Franchi Gianni
  • Yu Xuanlong
  • Bursuc Andrei
  • Tena Angel
  • Kazmierczak Rémi
  • Dubuisson Séverine
  • Aldea Emanuel
  • Filliat David
  , 2022. Predictive uncertainty estimation is essential for safe deployment of Deep Neural Networks in real-world autonomous systems. However, disentangling the different types and sources of uncertainty is non trivial for most datasets, especially since there is no ground truth for uncertainty. In addition, while adverse weather conditions of varying intensities can disrupt neural network predictions, they are usually under-represented in both training and test sets in public datasets.We attempt to mitigate these setbacks and introduce the MUAD dataset (Multiple Uncertainties for Autonomous Driving), consisting of 10,413 realistic synthetic images with diverse adverse weather conditions (night, fog, rain, snow), out-of-distribution objects, and annotations for semantic segmentation, depth estimation, object, and instance detection. MUAD allows to better assess the impact of different sources of uncertainty on model performance. We conduct a thorough experimental study of this impact on several baseline Deep Neural Networks across multiple tasks, and release our dataset to allow researchers to benchmark their algorithm methodically in adverse conditions. More visualizations and the download link for MUAD are available at https://muad-dataset.github.io/.
• Latent Discriminant deterministic Uncertainty
  
  • Franchi Gianni
  • Yu Xuanlong
  • Bursuc Andrei
  • Aldea Emanuel
  • Dubuisson Séverine
  • Filliat David
  , 2022. (10.1007/978-3-031-19775-8_15)
  DOI : 10.1007/978-3-031-19775-8_15
• Estimating the Coverage Measure and the Area Explored by a Side-Scan Sonar
  
  • Vianna Maria Luiza Costa
  • Goubault Eric
  • Jaulin Luc
  • Putot Sylvie
  , 2022, 2022, pp.1-6. Full coverage of an area of interest is a common task for a robot in the underwater environment. Estimating the area explored by the robot is indeed essential for determining if path-planning algorithms lead to complete coverage. In this work, we propose a method for estimating the area explored by a Side-Scan Sonar. The proposed method is able to determine how many times each portion of the space has been sensed by the sonar using a novel approach based on the topological properties of the environment that has been scanned, and more precisely on an estimation of certain winding numbers. This property is useful for localization inside homogeneous environments, e.g. the underwater environment, and assessment for potential revisiting missions. (10.1109/OCEANS47191.2022.9977121)
  DOI : 10.1109/OCEANS47191.2022.9977121
• On Monocular Depth Estimation and Uncertainty Quantification using Classification Approaches for Regression
  
  • Yu Xuanlong
  • Franchi Gianni
  • Aldea Emanuel
  , 2022. Monocular depth is important in many tasks, such as 3D reconstruction and autonomous driving. Deep learning based models achieve state-of-the-art performance in this field. A set of novel approaches for estimating monocular depth consists of transforming the regression task into a classification one. However, there is a lack of detailed descriptions and comparisons for Classification Approaches for Regression (CAR) in the community and no in-depth exploration of their potential for uncertainty estimation. To this end, this paper will introduce a taxonomy and summary of CAR approaches, a new uncertainty estimation solution for CAR, and a set of experiments on depth accuracy and uncertainty quantification for CAR-based models on KITTI dataset. The experiments reflect the differences in the portability of various CAR methods on two backbones. Meanwhile, the newly proposed method for uncertainty estimation can outperform the ensembling method with only one forward propagation. (10.1109/ICIP46576.2022.9897930)
  DOI : 10.1109/ICIP46576.2022.9897930
• An accelerated level-set method for inverse scattering problems
  
  • Audibert Lorenzo
  • Haddar Houssem
  • Liu Xiaoli
  SIAM Journal on Imaging Sciences, Society for Industrial and Applied Mathematics, 2022, 15 (3), pp.1576-1600. We propose a rapid and robust iterative algorithm to solve inverse acoustic scattering problems formulated as a PDE constrained shape optimization problem. We use a level-set method to represent the obstacle geometry and propose a new scheme for updating the geometry based on an adaptation of accelerated gradient descent methods. The resulting algorithm aims at reducing the number of iterations and improving the accuracy of reconstructions. To cope with regularization issues, we propose a smoothing to the shape gradient using a single layer potential associated with ik where k is the wave number. Numerical experiments are given for several data types (full aperture, backscattering, phaseless, multiple frequencies) and show that our method outperforms a non accelerated approach in terms of convergence speed, accuracy and sensitivity to initial guesses. (10.1137/21M1457783)
  DOI : 10.1137/21M1457783
• Etalement de spectre par séquence directe et constellations tournées pour communications tactiques
  
  • Arbi Tarak
  • Pasquero Oudomsack Pierre
  • Bazin Alexis
  • Geller Benoît
  , 2022. Les systèmes de communications sans fil sont vulnérables aux attaques par interférences car le média étant partagé, toute entité mal intentionnée peut facilement accéder au canal et envoyer des signaux de brouillage qui interfèrent avec le signal légitime et empêchent sa bonne réception. Les techniques à étalement de spectre par séquence directe (Direct-Sequence Spread Spectrum-DSSS) sont alors utiles pour limiter ce risque dans des situations critiques. Dans cet article, on propose d'utiliser conjointement le signal DSSS avec les constellations tournées. Les simulations, prenant en compte des phénomènes physiques comme un fort décalage Doppler, montrent que la méthode proposée permet un gain en robustesse de plusieurs dBs par rapport au DSSS utilisé avec des constellations conventionnelles.
• Numerical challenges in the simulation of 1D bounded low-temperature plasmas with charge separation in various collisional regimes
  
  • Reboul Louis
  • Massot Marc
  • Laguna Alejandro Alvarez
  , 2022, pp.190004. We study a 1D geometry of a plasma confined between two conducting floating walls with applications to laboratory plasmas. These plasmas are characterized by a quasi-neutral bulk that is joined to the wall by a thin boundary layer called sheath that is positively charged. Although analytical solutions are available in the sheath and the pre-sheath, joining the two areas by one analytical solution is still an open problem which requires the numerical resolution of the fluid equations coupled to Poisson equation. Current numerical schemes use high-order discretizations to correctly capture the electron current in the sheath, presenting unsatisfactory results in the boundary layer and they are not adapted to all the possible collisional regimes. In this work, we identify the main numerical challenges that arise when attempting the simulations of such configuration and we propose explanations for the observed phenomena via numerical analysis. We propose a numerical scheme with controlled diffusion as well as new discrete boundary conditions that address the identified issues. (10.1063/5.0187483)
  DOI : 10.1063/5.0187483
• A Simple Second-Order Implicit-Explicit Asymptotic Preserving Scheme for the Hyperbolic Heat Equations
  
  • Reboul Louis
  • Pichard Teddy
  • Massot Marc
  , 2022. We propose an asymptotic preserving (AP) Implicit-Explicit (ImEx) scheme for the hyperbolic heat equation in the diffusive regime. This scheme is second order in time and space and l ∞-stabile under relatively low constraints on the time step, and without requiring the use slope limiters. The construction exploits a formalism developed in a previous work and, compared to it, leads to a simpler implementation but it loses uniform accuracy on paper. Stability and accuracy are verified on numerical examples, and even uniform accuracy is obtained on those. Finally, we discuss extensions of this method to the non linear case of isothermal Euler equations with friction.
• Contact Line Catch Up by Growing Ice Crystals
  
  • Grivet Rodolphe
  • Monier Antoine
  • Huerre Axel
  • Josserand Christophe
  • Séon Thomas
  Physical Review Letters, American Physical Society, 2022, 128 (25), pp.254501. (10.1103/PhysRevLett.128.254501)
  DOI : 10.1103/PhysRevLett.128.254501
• Successive Convexification for Optimal Control with Signal Temporal Logic Specifications
  
  • Mao Yuanqi
  • Acikmese Behcet
  • Garoche Pierre-Loïc
  • Chapoutot Alexandre
  , 2022. As the scope and complexity of modern cyber-physical systems increase, newer and more challenging mission requirements will be imposed on the optimal control of the underlying unmanned systems. This paper proposes a solution to handle complex temporal requirements formalized in Signal Temporal Logic (STL) specifications within the Successive Convexification (SCvx) algorithmic framework. This SCvx-STL solution method consists of four steps: 1) Express the STL specifications using their robust semantics as state constraints. 2) Introduce new auxiliary state variables to transform these state constraints as system dynamics, by exploiting the recursively defined structure of robust STL semantics. 3) Smooth the resulting system dynamics with polynomial smooth min-and maxfunctions. 4) Convexify and solve the resulting optimal control problem with the SCvx algorithm, which enjoys guaranteed convergence and polynomial time subproblem solving capability. Our approach retains the expressiveness of encoding mission requirements with STL semantics, while avoiding the usage of combinatorial optimization techniques such as Mixed-integer programming. Numerical results are shown to demonstrate its effectiveness. (10.1145/3501710.3519518)
  DOI : 10.1145/3501710.3519518
• A benchmark of incremental model transformation tools based on an industrial case study with AADL
  
  • Mkaouar Hana
  • Blouin Dominique
  • Borde Etienne
  Software and Systems Modeling, Springer Verlag, 2022. (10.1007/s10270-022-00989-z)
  DOI : 10.1007/s10270-022-00989-z
• Investigation of the homogeneity of energy conversion processes at dipolarization fronts from MMS measurements
  
  • Alqeeq S. W.
  • Le Contel O.
  • Canu Patrick
  • Retinò Alessandro
  • Chust Thomas
  • Mirioni Laurent
  • Richard L.
  • Aït-Si-Ahmed Y.
  • Alexandrova A.
  • Chuvatin A.
  • Ahmadi N.
  • Baraka S. M.
  • Nakamura R.
  • Wilder F. D.
  • Gershman D. J.
  • Lindqvist P. A.
  • Khotyaintsev Yu. V.
  • Ergun R. E.
  • Burch J. L.
  • Torbert R. B.
  • Russell C. T.
  • Magnes W.
  • Strangeway R. J.
  • Bromund K. R.
  • Wei H.
  • Plaschke F.
  • Anderson B. J.
  • Giles B. L.
  • Fuselier S. A.
  • Saito Y.
  • Lavraud B.
  Physics of Plasmas, American Institute of Physics, 2022, 29 (1), pp.012906. We report on six dipolarization fronts (DFs) embedded in fast earthward flows detected by the Magnetospheric Multiscale mission during a substorm event on 23 July 2017. We analyzed Ohm’s law for each event and found that ions are mostly decoupled from the magnetic field by Hall fields. However, the electron pressure gradient term is also contributing to the ion decoupling and likely responsible for an electron decoupling at DF. We also analyzed the energy conversion process and found that the energy in the spacecraft frame is transferred from the electromagnetic field to the plasma (J E > 0) ahead or at the DF, whereas it is the opposite (J E < 0) behind the front. This reversal is mainly due to a local reversal of the cross-tail current indicating a substructure of the DF. In the fluid frame, we found that the energy is mostly transferred from the plasma to the electromagnetic field (J E0 < 0) and should contribute to the deceleration of the fast flow. However, we show that the energy conversion process is not homogeneous at the electron scales due to electric field fluctuations likely related to lower-hybrid drift waves. Our results suggest that the role of DF in the global energy cycle of the magnetosphere still deserves more investigation. In particular, statistical studies on DF are required to be carried out with caution due to these electron scale substructures. (10.1063/5.0069432)
  DOI : 10.1063/5.0069432
• RINO: Robust INner and Outer Approximated Reachability of Neural Networks Controlled Systems
  
  • Goubault Eric
  • Putot Sylvie
  Lecture Notes in Computer Science, Springer, 2022, pp.511 - 523. We present a unified approach, implemented in the RINO tool, for the computation of inner and outer-approximations of reachable sets of discrete-time and continuous-time dynamical systems, possibly controlled by neural networks with differentiable activation functions. RINO combines a zonotopic set representation with generalized mean-value AE extensions to compute under and over-approximations of the robust range of differentiable functions, and applies these techniques to the particular case of learning-enabled dynamical systems. The AE extensions require an efficient and accurate evaluation of the function and its Jacobian with respect to the inputs and initial conditions. For continuous-time systems, possibly controlled by neural networks, the function to evaluate is the solution of the dynamical system. It is over-approximated in RINO using Taylor methods in time coupled with a set-based evaluation with zonotopes. We demonstrate the good performances of RINO compared to state-of-the art tools Verisig 2.0 and ReachNN* on a set of classical benchmark examples of neural network controlled closed loop systems. For generally comparable precision to Verisig 2.0 and higher precision than ReachNN*, RINO is always at least one order of magnitude faster, while also computing the more involved inner-approximations that the other tools do not compute. (10.1007/978-3-031-13185-1_25)
  DOI : 10.1007/978-3-031-13185-1_25
• Direct Quantitative Characterization of Polymer Brushes obtained by Surface-Initiated ATRP on Silicon
  
  • Gouget-Laemmel Anne-Chantal
  • Zidelmal Nacim
  • Soares Rafaela S B
  • Aubry-Barroca Nadine
  • Dragoe Diana
  • Lepoittevin Bénédicte
  • Salmi-Mani Hanène
  • Mellah Mohamed
  • Henry-De-Villeneuve Catherine
  • Ozanam François
  • Schulz Emmanuelle
  • Roger Philippe
  ACS Applied Polymer Materials, American Chemical Society, 2022, 5 (1), pp.517–528. With respect to the increasing need for fully characterizing surface-tethered polymer brushes, the capacity of quantitative IR-Fourier transform infrared (FTIR) spectroscopy using a multiple-internal-reflection Si prism as the attenuated total reflection (ATR) element is investigated to directly characterize the surface chemical modifications occurring during a surface-initiated controlled polymerization. A simple two-step strategy is used involving first the covalent grafting of atom transfer radical polymerization (ATRP) initiators on a hydrogenated silicon surface and the subsequent polymerization of styrene. Three prefunctionalized surfaces designated Si-Br1, Si-Br2, and Si-Br3 are obtained by different procedures. The initiator grafting densities obtained by quantitative IR are 1.7 ± 0.3 nm–2 for Si-Br1, 1.5 ± 0.3 nm–2 for Si-Br2, and 0.9 ± 0.2 nm–2 for Si-Br3. After the polymerization of styrene under the same experimental conditions (grafting from without sacrificial initiators) and a careful Soxhlet rinse to remove physisorbed polymers formed in solution, almost no polymerization is observed using Si-Br1 with a value of the density in polymerized styrene units of 12 ± 2 nm–2, which is probably due to the chelating effect of the amino linkers used for grafting the initiators in Si-Br1. In contrast, the densities in styrene units are 54 ± 11 nm–2 using Si-Br2 and 141 ± 28 nm–2 using Si-Br3. The degree of polymerization (DP) has been evaluated by measuring the polymer thickness (by ellipsometry and atomic force microscopy, AFM) and using a scaling law relating the latter to DP for dry polymer brushes. High DP values of 200 and 1000 are found in the case of Si-Br2 and Si-Br3, respectively. The fraction of active polymerization initiators is found to be 15–18%, independent of the initiator surface density. In contrast, polymerization kinetics appear affected by steric hindrance and conformational disorder among grafted initiators. This approach for determining surface densities of grafted initiators and grafted polymer chains and DPs is fully generalizable to any other polymer system. (10.1021/acsapm.2c01632)
  DOI : 10.1021/acsapm.2c01632