Search

This database outlines the development of a numerical model for simulating pavement mechanical behavior under the Traffic Speed Deflectometer (TSD).

Successfully addressing a wide variety of tasks is a core ability of autonomous agents, requiring flexibly adapting the underlying decision-making strategies and, as we argue in this work, also adapting the perception modules. An analogical argument would be the human visual system, which uses top-down signals to focus attention determined by the current task. Similarly, we adapt pre-trained large vision models conditioned on specific downstream tasks in the context of multi-task policy learning. We introduce task-conditioned adapters that do not require finetuning any pre-trained weights, combined with a single policy trained with behavior cloning and capable of addressing multiple tasks. We condition the visual adapters on task embeddings, which can be selected at inference if the task is known, or alternatively inferred from a set of example demonstrations. To this end, we propose a new optimization-based estimator. We evaluate the method on a wide variety of tasks from the CortexBench benchmark and show that, compared to existing work, it can be addressed with a single policy. In particular, we demonstrate that adapting visual features is a key design choice and that the method generalizes to unseen tasks given a few demonstrations.

Gas-particle flows are commonly simulated through two-fluid model at industrial-scale. However, these simulations need very fine grid to have accurate flow predictions, which is prohibitively demanding in terms of computational resources. To circumvent this problem, the filtered two-fluid model has been developed, where large-scale flow field is numerically resolved and small-scale fluctuations are accounted for through subgrid-scale modeling. In this study, we have performed fine-grid two-fluid simulations of dilute gas-particle flows in periodic domains and applied explicit filtering to generate datasets. Then, these datasets have been used to develop artificial neural network (ANN) models for closures such as the filtered drag force and solid phase stress for the filtered two-fluid model. The set of input variables for the subgrid drag force ANN model that has been found previously to work well for dense flow regimes is found to work as well for the dilute regime. In addition, we present a Galilean invariant tensor basis neural network (TBNN) model for the filtered solid phase stress which can capture nicely the anisotropic nature of the solid phase stress arising from subgrid-scale velocity fluctuations. Finally, the predictions provided by this new TBNN model are compared with those obtained from a simple eddy-viscosity ANN model.

Among the commonly used non-destructive techniques, the Ground Penetrating Radar (GPR) is one of the most widely adopted today for assessing pavement conditions in France. However, conventional radar systems and their forward processing methods have shown their limitations for the physical and geometrical characterization of very thin layers such as tack coats. However, the use of Machine Learning methods applied to GPR with an inverse approach showed that it was numerically possible to identify the tack coat characteristics despite masking effects due to low timefrequency resolution noted in the raw B-scans. Thus, we propose in this paper to apply the inverse approach based on Machine Learning, already validated in previous works on numerical data, on two experimental cases with different pavement structures. The first case corresponds to a validation on known pavement structures on the Gustave Eiffel University (Nantes, France) with its pavement fatigue carousel and the second case focuses on a new real road in Vend{\'e}e department (France). In both case studies, the performances of SVM/SVR methods showed the efficiency of supervised learning methods to classify and estimate the emulsion proportioning in the tack coats.

In numerical weather prediction (NWP), a large number of observations are used to create initial conditions for weather forecasting through a process known as data assimilation. An assessment of the value of these observations for NWP can guide us in the design of future observation networks, help us to identify problems with the assimilation system, and allow us to assess changes to the assimilation system. However, the assessment can be challenging in convection-permitting NWP. First, the strong nonlinearity in the forecast model limits the methods available for the assessment. Second, convection-permitting NWP typically uses a limited area model and provides short forecasts, giving problems with verification and our ability to gather sufficient statistics. Third, convection-permitting NWP often makes use of novel observations, which can be difficult to simulate in an observing system simulation experiment (OSSE). We compare methods that can be used to assess the value of observations in convection-permitting NWP and discuss operational considerations when using these methods. We focus on their applicability to ensemble forecasting systems, as these systems are becoming increasingly dominant for convection-permitting NWP. We also identify several future research directions: comparison of forecast validation using analyses and observations, the effect of ensemble size on assessing the value of observations, flow-dependent covariance localization, and generation and validation of the nature run in an OSSE., Comment: 35 pages

This paper addresses the theoretical foundations of pedestrian models for crowd dynamics. While the topic gains momentum, current models differ widely in their mathematical structure, even if we only consider continuous agent-based models. To clarify their underpinning, we first lay mathematical foundations for the common hierarchical decomposition into strategic, tactical, and operational levels and underline the practical interest in preserving the continuity between the latter two levels by working with a floor field, rather than way-points. Turning to local navigation, we clarify how three archetypical approaches, namely, purely reactive models, anticipatory models based on the idea of times to collision, and game theory, differ in the way they extrapolate trajectories in the near future. We also insist on the oft-overlooked distinction between processes pertaining to decision-making and mechanical effects in dense settings. The implications of these differences are illustrated with a comparison of the numerical predictions of these models in the simple scenario of head-on collision avoidance between agents, by varying the walking speed, the reaction times, and the degree of courtesy of the agents, notably.

Distributed infrastructures for computation and analytics are now evolving towards an interconnected ecosystem allowing complex scientific workflows to be executed across hybrid systems spanning from IoT Edge devices to Clouds, and sometimes to supercomputers (the Computing Continuum). Understanding the performance trade-offs of large-scale workflows deployed on such complex Edge-to-Cloud Continuum is challenging. To achieve this, one needs to systematically perform experiments, to enable their reproducibility and allow other researchers to replicate the study and the obtained conclusions on different infrastructures. This breaks down to the tedious process of reconciling the numerous experimental requirements and constraints with low-level infrastructure design choices.To address the limitations of the main state-of-the-art approaches for distributed, collaborative experimentation, such as Google Colab, Kaggle, and Code Ocean, we propose KheOps, a collaborative environment specifically designed to enable cost-effective reproducibility and replicability of Edge-to-Cloud experiments. KheOps is composed of three core elements: (1) an experiment repository; (2) a notebook environment; and (3) a multi-platform experiment methodology.We illustrate KheOps with a real-life Edge-to-Cloud application. The evaluations explore the point of view of the authors of an experiment described in an article (who aim to make their experiments reproducible) and the perspective of their readers (who aim to replicate the experiment). The results show how KheOps helps authors to systematically perform repeatable and reproducible experiments on the Grid5000 + FIT IoT LAB testbeds. Furthermore, KheOps helps readers to cost-effectively replicate authors experiments in different infrastructures such as Chameleon Cloud + CHI@Edge testbeds, and obtain the same conclusions with high accuracies (> 88% for all performance metrics).

Implicit representations such as Neural Radiance Fields (NeRF) have been shown to be very effective at novel view synthesis. However, these models typically require manual and careful human data collection for training. In this paper, we present AutoNeRF, a method to collect data required to train NeRFs using autonomous embodied agents. Our method allows an agent to explore an unseen environment efficiently and use the experience to build an implicit map representation autonomously. We compare the impact of different exploration strategies including handcrafted frontier-based exploration, end-to-end and modular approaches composed of trained high-level planners and classical low-level path followers. We train these models with different reward functions tailored to this problem and evaluate the quality of the learned representations on four different downstream tasks: classical viewpoint rendering, map reconstruction, planning, and pose refinement. Empirical results show that NeRFs can be trained on actively collected data using just a single episode of experience in an unseen environment, and can be used for several downstream robotic tasks, and that modular trained exploration models outperform other classical and end-to-end baselines. Finally, we show that AutoNeRF can reconstruct large-scale scenes, and is thus a useful tool to perform scene-specific adaptation as the produced 3D environment models can be loaded into a simulator to fine-tune a policy of interest.

The collection of electrical consumption time series through smart meters grows with ambitious nationwide smart grid programs. This data is both highly sensitive and highly valuable: strong laws about personal data protect it while laws about open data aim at making it public after a privacy-preserving data publishing process. In this work, we study the uniqueness of large scale real-life fine-grained electrical consumption time-series and show its link to privacy threats. Our results show a worryingly high uniqueness rate in such datasets. In particular, we show that knowing 5 consecutive electric measures allows to re-identify on average more than 90% of households in our 2.5M half-hourly electric time series dataset. Moreover, uniqueness remains high even when data is severely degraded. For example, when data is rounded to the nearest 100 watts, knowing 7 consecutive electric measures allows to re-identify on average more than 40% of the households (same dataset). We also study the relationship between uniqueness and entropy, uniqueness and electric consumption, and electric consumption and temperatures, showing their strong correlation.

Understanding and mapping a new environment are core abilities of any autonomously navigating agent. While classical robotics usually estimates maps in a stand-alone manner with SLAM variants, which maintain a topological or metric representation, end-to-end learning of navigation keeps some form of memory in a neural network. Networks are typically imbued with inductive biases, which can range from vectorial representations to birds-eye metric tensors or topological structures. In this work, we propose to structure neural networks with two neural implicit representations, which are learned dynamically during each episode and map the content of the scene: (i) the Semantic Finder predicts the position of a previously seen queried object; (ii) the Occupancy and Exploration Implicit Representation encapsulates information about explored area and obstacles, and is queried with a novel global read mechanism which directly maps from function space to a usable embedding space. Both representations are leveraged by an agent trained with Reinforcement Learning (RL) and learned online during each episode. We evaluate the agent on Multi-Object Navigation and show the high impact of using neural implicit representations as a memory source.

This paper explores an integrated approach for improved sensing and positioning with applications in air traffic management (ATM) and in the Advanced Surface Movement Guidance and Control System (A-SMGCS). The integrated approach includes the synergy of 3D Vector Antenna with the novel time-of-arrival and angle-of-arrival estimate methods for accurate positioning, combining the sensing on the sub-6GHz and mmWave spectrum for the enhanced non-cooperative surveillance. For the positioning scope, both uplink and downlink 5G reference signals are investigated and their performance is evaluated. For the non-cooperative sensing scope, a novel 5G-signal-based imaging function is proposed and verified with realistic airport radio-propagation modelling and the AI-based targets tracking-and-motion recognition are investigated. The 5G-based imaging and mmWave radar based detection can be potentially fused to enhance surveillance in the airport. The work is being done within the European-funded project NewSense and it delves into the 5G, Vector Antennas, and mmWave capabilities for future ATM solutions., Comment: 8 pages, 15 figures

We address planning and navigation in challenging 3D video games featuring maps with disconnected regions reachable by agents using special actions. In this setting, classical symbolic planners are not applicable or difficult to adapt. We introduce a hybrid technique combining a low level policy trained with reinforcement learning and a graph based high level classical planner. In addition to providing human-interpretable paths, the approach improves the generalization performance of an end-to-end approach in unseen maps, where it achieves a 20% absolute increase in success rate over a recurrent end-to-end agent on a point to point navigation task in yet unseen large-scale maps of size 1km x 1km. In an in-depth experimental study, we quantify the limitations of end-to-end Deep RL approaches in vast environments and we also introduce "GameRLand3D", a new benchmark and soon to be released environment can generate complex procedural 3D maps for navigation tasks.

We present Godot Reinforcement Learning (RL) Agents, an open-source interface for developing environments and agents in the Godot Game Engine. The Godot RL Agents interface allows the design, creation and learning of agent behaviors in challenging 2D and 3D environments with various on-policy and off-policy Deep RL algorithms. We provide a standard Gym interface, with wrappers for learning in the Ray RLlib and Stable Baselines RL frameworks. This allows users access to over 20 state of the art on-policy, off-policy and multi-agent RL algorithms. The framework is a versatile tool that allows researchers and game designers the ability to create environments with discrete, continuous and mixed action spaces. The interface is relatively performant, with 12k interactions per second on a high end laptop computer, when parallized on 4 CPU cores. An overview video is available here: https://youtu.be/g1MlZSFqIj4