Archive ouverte HAL-LAAS
Not a member yet
12189 research outputs found
Sort by
When Quality Matters: Constraint Programming for Automated Temporal and Numeric Planning
No full textInternational audienceAutomated planning is a field of Artificial Intelligence interested in finding a set of actions that drives the evolution of the environment from an initial state to a desired goal state. Much of the work of the community has been on socalled domain-independent planning where a solver is expected to produce a plan from an abstract problem, specified in a common description language. This has led the community to produce a number of highly-efficient solvers that can be expected to work on a large variety of domains without any fine-tuning.Where most of the work has focused on sequential plans over purely symbolic states, we instead propose a constraint-based planner whose focus is on more expressive variants, namely numeric and temporal planning, essential in many practical applications. We extend an existing CP encoding of temporal planning with support for optimization and numeric states and leverage an existing lazy clause generation CP solver to find and optimize plans. Where the most successful automated planners rely on some form of forward-search, we show that constraintprogramming can be just as effective in finding satisfiable solutions while substantially improving the quality of the produced plan
Hybrid Lyapunov-based feedback stabilization of bipedal locomotion based on reference spreading
No full textInternational audienceWe propose a hybrid formulation of the linear inverted pendulum model for bipedal locomotion, where the foot switches are triggered based on the center of mass position, removing the need for pre-defined footstep timings. Using a concept similar to reference spreading, we define nontrivial tracking error coordinates induced by our hybrid model. These coordinates enjoy desirable linear flow dynamics and rather elegant jump dynamics perturbed by a suitable extended class function of the position error. We stabilize this hybrid error dynamics using a saturated feedback controller, selecting its gains by solving a convex optimization problem. We prove local asymptotic stability of the tracking error and provide a certified estimate of the basin of attraction, comparing it with a numerical estimate obtained from the integration of the closed-loop dynamics. Simulations on a full-body model of a real robot show the practical applicability of the proposed framework and its advantages with respect to a standard model predictive control formulation
Fractional Digital Regenerative Frequency Dividers
No full textInternational audienceABSTRACT This paper describes the topology and operation of a new family of asynchronous frequency dividers called fractional digital regenerative frequency dividers. The operation is explained using a digital approach in the time domain compared to the more classical analogue approach in the frequency domain. A fractional behaviour is intrinsic to this topology without the mandatory need of an external modulator. Integer ratios can also be easily obtained, for example to serve as multi‐modulus dividers when required. Starting with the elementary version of this kind of divider, a generalization is made with repeated cell patterns in the closed loop for finer control of the fractional ratio. Some measurements on an FPGA‐based implementation are provided to validate the division ratio of the elementary version as well as an example of a more complex configuration
Robust RL Navigation Via Sensitivity-Aware Observation Augmentation
No full textInternational audienceIn this work, we propose Observation Augmentation (OA), an approach that leverages model-based sensitivity analysis to adapt the observation of a baseline reinforcement learning policy that improves robustness to uncertainties without requiring tedious training environment tuning or retraining. We demonstrate this method on a quadrotor gatepassing task, where the OA method improves the safety of the baseline method under observation uncertainties. Furthermore, this method is highly efficient and flexible, and can be easily applied to any robot system
TCAD simulations of a barrier structure designed to improve the performance of very-low bandgap InAs/InAsSb thermophotovoltaic cells
No full textInternational audienceVery-low bandgap TPV cells (< 0.4 eV) usually require cryogenic cooling (∼ 70 K) because of their high intrinsic dark current densities. To mitigate this effect, a barrier structure combined with a gallium-free InAs/InAsSb absorber, inspired by infrared photodetectors, is studied through optoelectronic TCAD simulations. This analysis reveals that the power output of the cell could be increased by 24% at 200 K thanks to this design, when compared to an equivalent PIN structure. The quantitative effects of contact doping, parasitic absorption, and carrier diffusion length are discussed in detail. An optimum design is extracted, predicting a power output of 0.3 W/cm2 and pairwise efficiency of 16% at 200 K under the illumination of an 800 °C black-body emitter. The physical mechanisms limiting the performance of the cell at higher temperatures are identified, and a guideline for future improvements is proposed
Opto-electrical model of a single-junction photovoltaic cell using Monte Carlo methods
No full textInternational audienceIn the context of the development of solar photovoltaics for global energy production, it remains essential to clearly identify performance limitations. The present article revisits the interpretations of photovoltaic cell performance by introducing a path-integral formulation of the current. The case of a single p-n junction cell operating in low injection conditions is examined as a proof of concept. The total current is calculated by implementing two different algorithms, which exhibit two viewpoints for analyzing the physics involved. The first one describes random walks on a reverse path space, back to the source of charge carriers. The second one describes random walks on a direct path space, starting from the source. Both algorithms are validated against an analytical solution. A Monte Carlo algorithm is also proposed for directly determining the voltage at the maximum power point without requiring the entire current-voltage curve. The calculation of a spectral-spatial internal quantum efficiency, combined with a statistical analysis of the random paths, demonstrates the relevance of this approach for interpreting the involved physics. Numerous benefits are anticipated from our advancements in Monte Carlo techniques: enhanced physics representations, insensitivity of computation time to the dimension of the integration domain, and to the geometrical complexity at all scales
PECVD of SiON optical thin films and nano-laminates
No full textInternational audienceHere, we report a method to create optical thin films and nano-laminates [5] of mixed compounds of silica (SiOx), silicon nitride (SiNy) and amorphous silicon (a-Si) by Inductive-Coupled-Plasma PECVD and demonstrate their use to fabricate anti-reflective coatings, optical mirrors, as well as birefringent structures
Contribution à la définition des enjeux et verrous de la soutenabilité en électronique de puissance
No full textInternational audienceThe GT-CEPPS (Working Group on More Sustainable Electronic Power Converters) of the GDR SEEDS (Electrical Energy Systems in their Societal Dimensions) presents here the results of its work on the challenges and obstacles to sustainability in power electronics. The work focuses on two aspects of the topic: tools, methods and indicators on the one hand, and circular design on the other. The analysis is structured into four levels corresponding to degrees of difficulty, transformation and societal acceptability. At each level, the challenges and obstacles are described and then translated into avenues for research activities. The whole is contextualised and illustrated through ongoing activities at national and European level.Le GT-CEPPS (Groupe de travail -Convertisseurs Electroniques de Puissance Plus Soutenables) du GDR SEEDS (Systèmes d'Energie Electrique dans leur Dimensions Sociétales) présente ici le fruit de son travail de réflexion autour des enjeux et verrous de la soutenabilité en électronique de puissance. Le travail porte sur deux facettes de la thématique, les outils, méthodes, indicateurs d'une part, et la conception en vue de la circularité d'autre part. L'analyse est structurée en quatre niveaux qui correspondent à des échelons de difficulté, de transformation et d'acceptabilité sociétale. Dans chaque niveau, des enjeux et des verrous sont décrits et ensuite traduits en pistes d'activités de recherche. L'ensemble est contextualisé et illustré via les activités en cours aux échelles nationale et européenne
An evaluation of Sample Average Approximation applied to the design of impulsive thrust space collision avoidance maneuvers
No full textInternational audienceThis paper investigates how Sample Average Approximation (SAA) techniques can be applied tochance-constrained optimization problems defined for collision avoidance maneuvers design betweenan active satellite and a passive space debris. The objective is to compute impulsive evasive maneuversthat minimize the fuel consumption of the active asset and simultaneously maintaining the collisionprobability below a given threshold. A particular SAA algorithm is then proposed and compared againsta direct optimization algorithm using both synthetic and real conjunction data
Complete Upper Bound Hierarchies for Spectral Minimum in Noncommutative Polynomial Optimization
No full textThis is the full/expanded/comprehensive version of the preliminary report https://laas.hal.science/hal-04440949v1This work addresses the problem of computing the spectral minimum (ground state energy) of a noncommutative polynomial subject to finitely many noncommutative polynomial constraints.Building on the Helton-McCullough Positivstellensatz, the Navascués-Pironio-Acı́n (NPA) hierarchy provides a sequence of lower bounds that converge to the spectral minimum under mild assumptions on the constraint set. Each of these bounds can be computed via semidefinite programming.In this paper, we develop complementary, complete hierarchies of upper bounds for the spectral minimum. These are noncommutative counterparts to Lasserre’s upper bound hierarchies for polynomial optimization. Each upper bound is obtained by solving a generalized eigenvalue problem. The proposed hierarchies are applicable to optimization problems in both bounded and unbounded operator algebras, as illustrated through a range of examples