Search

The paper presents a complex simulation system for demonstrating the evacuation process in a building, whereby people attempt to escape from a dangerous scenario. It is novel in that it integrates a range of different models: agent-based model, social force model, and psychological behaviour with emotions and norms. The method uses the communication network based on the message queuing telemetry transport protocol that assists to gather information from the environment. The paths are modified using feelings and rule-based expert system. The authors conduct some simulations and conclude with recommendations for management of safer environments.

Weather forecasting is the task to predict the state of the atmosphere in a given location. In the past, the weather forecast has been done through physical models of the atmosphere as a fluid. It becomes the problem of solving sophisticated equations of fluid dynamics. In recent years, machine learning algorithms have been used to speed up weather data modeling, a computationally intensive task. Machine learning algorithms learn from data and produce relevant predictions. In addition to prediction, there is a need of providing knowledge about domain relationships inside the data. This paper provides a new approach using interpretable machine learning for explaining the characteristic variables of meteorological data. Interpretable machine learning is the use of machine learning models for the extraction of knowledge in the data. An illustration is shown on characteristic variables of meteorological data.

On-board restoration of underwater images on embedded platforms such as marine drones is faced with many obstacles including image quality and real-time constraints. Confrontations met on the way vary from views to solution methods depending on the goals, such as underwater vehicle realtime control and positioning or underwater object recognition. In this research, five algorithms for underwater image restoration were studied and evaluated. In order to evaluate the quality of the performing algorithms nine evaluation criteria were used. Split into two types, the no reference metrics assesses only the quality of the image results, while the full reference criterion uses a reference image to estimate it. The calculation of these criteria allows each algorithm to be compared. Furthermore, the possibility to optimise algorithms in order to make them applicable to meet real-time requirements on embedded platforms was investigated., La restauration embarquée d'images sous-marines sur des plates-formes embarquées telles que les drones marins est confrontée à de nombreux obstacles dont la qualité des images et les contraintes de temps réel. Les confrontations rencontrées en cours de route varient des points de vue aux méthodes de solution en fonction des objectifs, tels que le contrôle et le positionnement en temps réel du véhicule sous-marin ou la reconnaissance d'objets sous-marins. Dans cette recherche, cinq algorithmes de restauration d'images sous-marines ont été étudiés et évalués. Afin d'évaluer la qualité des algorithmes performants, neuf critères d'évaluation ont été utilisés. Divisés en deux types, le critère sans référence évalue uniquement la qualité des résultats de l'image, tandis que le critère avec référence complète utilise une image de référence pour l'estimer. Le calcul de ces critères permet de comparer chaque algorithme. En outre, la possibilité d'optimiser les algorithmes afin de les rendre applicables pour répondre aux exigences du temps réel sur les plateformes embarquées a été étudiée.

International audience; Division of labour is a key aspect of distributed systems, such as swarm robotics or multi-agent systems. Inspired by social insects known for their task allocation capabilities, most of the models rely on two assumptions: 1) each task is associated with a stimulus, and 2) the execution of this task lowers that stimulus. In short, the stimulus is a representation of the amount of work needed on a task. When these assumptions are not true, we need a mechanism to guide the agent in its decision whether to pursue or to interrupt its current task, as there is no diminishing stimulus to rely on. In this article, we propose a model based on the Response Threshold Model and a mechanism based on the agent’s intrinsic motivation and internal states, allowing to take into account tasks dissociated from stimuli. Agents use their intrinsic motivation to emulate the priority of tasks not associated with any stimuli, and to decide whether to interrupt or pursue their current task. This model has been applied to simulate the division of labour within a simplified honey bee colony, associated with the constantly adapting physiology of honey bees. Preliminary results show that the task allocation is effective, robust and in some cases improved by the interruption mechanism.

Viability is a very important feature of dynamic systems under state constraints whose initial value problem does not ensure uniqueness of solutions. In this paper, we introduce an hybrid automaton to address the question of viability of a cellular tissue. This hybrid automaton couples two dynamical models: differential equations manage the energy of the system and morphological equations govern the growth of the tissue. The cells can proliferate when they have enough access to oxygen and nutrient to produce the energy, remain quiescent when this energy is between two levels, or die when this energy is too low. The constraint we choose is to maintain the number of cells of the tissue during a certain time horizon. We have shown that for all the 1029 2D-tissues of 16 cells with an associate genotype, only 5 are viable for this constraint in a long time horizon. Moreover, for all these tissues, they renew there cells periodically. These periodic shapes are like periodic limit cycles in the state space of shapes.

The goal of this work is to introduce a mathematical model of multicellular developmental design based on morphological analysis in order to study the robustness of multi-cellular organism development. In this model each cell is a controlled system and has the same information, an ordered list of cell type. Cells perceive their neighbours during the growth process and decide to divide in a direction given by the reading advancement of the virtual genetic material and depending on the complex interplay between genetic, epigenetic and environment. Cells can perform distinct functions but in our simulator, two cell types just differ by there color and by permuting the segmentation direction according to the virtual genetic material and the epigenetic control. The switching on and switching off of genes depends on the environment of the cell.The multi-cellular organism has to reach a shape in a given environment to which it has to adapt. We present in this paper an algorithm based model which is implemented in a virtual 3D-environment. Moreover, the algorithm follows the principle of inertia in that the cells progress through the reading of its virtual genetic material after a punctuated equilibrium or when its viability is at stake.

International audience; Social insects and more specifically honey bees have very complex, powerful and interesting task allocation abilities. They are able to distribute their workforce effectively without any central control, using simple mechanisms based on stimuli (dances, pheromones), interactions, thresholds and feedback loops. Self-organisational concepts and some of those mechanisms, like pheromones, are invisible to the human eye. In order to help the user grasp the complexity of the task allocation, we propose to make them visible in a virtual, immersive and interactive environment. First, this implies that the system must be able to simulate and display in real-time around 30 000 bees interacting with each other, emitting clouds of pheromones. Secondly, this implies also that the user should be able to alter in real-time the environment of the bees (e.g. by manipulating the frame of the hive) and visualise the effects on the organisation, potentially days later. Finally, we would like to give to biologists and beekeepers some domain-related, intuitive and natural ways of interacting with the hive. We describe in this article these issues in more details, and how we plan to tackle them. This is a "work in progress", therefore a lot of work has still to be done, mostly surveying and modelling the interactions.

In this article, we use the potential of computational biology to highlight the key role of cell apoptosis for studying some tissue's properties through in silico experiments of morphogenesis. Our morphogenesis model is a new approach focusing on the deterministic program within cells that controls their placement and their differentiation at the beginning of the embryogenesis. Indeed, when the tissue is made by just a few pair of cells, we consider that cellular mechanisms are related neither to the influence of mechanical forces nor to the spread of chemicals. Dynamics are based on spatial and logical choices, the other factors being involved when the tissue contains a large number of cells. We had established a mathematical formulation of such a model and had enlightened the link between phenotype (cell placement and cell differentiation) and genotype (cell program) at the early embryogenesis. Indeed, that work allowed for generating any early tissue and the associated program that designs it. We propose now to study and assess some properties of these tissues for further selection and classification purposes. More precisely, we present in this article novel methods to measure tissue robustness based on the backward morphogenesis of our model. We also show some implementations of their self-maintenance properties, on the one hand to deal with environment disturbances through autopoiesis and on the other hand to achieve a dynamical steady state which ensures tissue renewal.

We present a multi-scale approach of tumor modeling in order to predict its evolution during radiotherapy. Within this context we focus on three different scales of tumor modeling: microscopic (individual cells in a voxel), mesoscopic (population of cells in a voxel) and macroscopic (whole tumor), with transition interfaces between these three scales. At the cellular level, the description is based on phase transfer probabilities in the cellular cycle. At the mesoscopic scale we represent populations of cells according to different stages in a cell cycle. Finally, at the macroscopic scale, the tumor description is based on the use of FDG PET image voxels. These three scales exist naturally: biological data are collected at the macroscopic scale, but the pathological behavior of the tumor is based on an abnormal cell-cycle at the microscopic scale. On the other hand, the introduction of a mesoscopic scale is essential in order to reduce the gap between the two extreme, in terms of resolution, description levels. It also reduces the computational burden of simulating a large number of individual cells. As an application of the proposed multi-scale model, we simulate the effect of oxygen on tumor evolution during radiotherapy. Two consecutive FDG PET images of 17 rectal cancer patients undergoing radiotherapy are used to simulate the tumor evolution during treatment. The simulated results are compared with those obtained on a third FDG PET image acquired two weeks after the beginning of the treatment.

International audience; The agent-based approach has been successfully used in the past years to model and simulate complex systems. We use this approach on a honeybee colony in a Dadant hive, where several tens of thousands of bees interact, in order to evaluate the impact of local actions at the bee-level (such as beekeeping practices) on the global system. In this article, we focus on the foraging activity, its recruitment mechanisms and the behaviour of foraging bees, and how these bees interact with the hive's environment, greatly different in scale. We present a customizable, agent-compliant module called the Ecosystem Module, that aims at modelling ant simulating the foraging, according to the local weather and the surrounding nectar sources. First results back up our model, showing that these recruitment mechanisms lead to a self-organizing process of the best available sources selection by the agents.

Nowadays information technology and communication has evolve, computer networks became vulnerable faced to new forms of threats. In this article, a new model of intrusion detection based on multi-agents system and inspired from the biological immune system is presented.We begin through a presentation of the biological immune systems, followed by immune algorithm, a model of artificial immune system which is integrated in the behavior of distributed agents on the network is proposed in order to ensure a good intrusions detection. The internal structure of the immune agents and their capacity to distinguish between self and not-self is also presented. Agents are able to achieve simultaneous treatments, they are auto-adaptable to environment evolution and have also the property of distributed coordination.

Modelling and teaching complex biological systems is a difficult process. Multi-Agent Based Simulations (MABS) have proved to be an appropriate approach both in research and education when dealing with such systems including emergent, self-organizing phenomena. This chapter presents NetBioDyn, an original software aimed at biologists (students, teachers, researchers) to easily build and simulate complex biological mechanisms observed in multicellular and molecular systems. Thanks to its specific graphical user interface guided by the multi-agent paradigm, this software does not need any prerequisite in computer programming. It thus allows users to create in a simple way bottom-up models where unexpected behaviours can emerge from many interacting entities. This multi-platform software has been used in middle schools, high schools and universities since 2010. A qualitative survey is also presented, showing its ability to adapt to a wide and heterogeneous audience. The Java executable and the source code are available online at http://virtulab.univ-brest.fr.

Nowadays the basic graphic cards contain more than 300 compute cores, which is expected to significantly increase the performances of parallel applications. In the past few years, several works were made concerning the simulation of Multi-agent Systems on the Graphics Processing Unit GPU. But they often use agents which have a position in the environment associated with an interaction radius. In this study we present a generic implementation on the GPU of reactive agents, which are formed of a set of positions. The communications between the agents are only indirect, and are possible by reading the trail left by other agents. In our system these trails diffuse in the environment. The problem posed is to compute the diffusion in an efficient way, otherwise the whole environment should be iterated. Our implementation takes advantage of the GPU by executing each agent and diffusion on one compute unit. Firstly, our implementation keeps the same dynamics in comparison to previous studies, where the agents were not executed in parallel. Secondly, the simulation time has a 45 × speed advantage over a 4 core CPU.

We present a new approach to understand shape emergence in a multicellular system. We set the hypothesis that beyond the influence of mechanical forces and gene expression, spatial constraints applied to the cells over time play a key role, mainly at the early stage of the embryo. In our model, cells are elements of a 2D cellular automaton-oriented multiagent system. To achieve a specific shape, they read actions within a genetic program according to the current timestep and logical constraints. This model has been used to generate effectively all possible tissue phenotypes (shapes) at any stage of the early embryo and their associated genotypes (genetic programs). We also build a model that couples the morphological equation governing the growth of the phenotypes with differential equations that diffuse energy in the system. This model has been used to do some pattern predictions, to define properties on tissues for the purpose of classification, and to simulate responses to therapy in the case of pathological tissues.

The cellular Potts model (CPM) is a cellular automaton (CA) for modelling morphogenesis. It characterizes a cell by a volume, a surface and a type. We have extended this model to add the notion of energetic exchange between cells via the secretion and consumption of molecules. These molecules are transformed into energy, allowing the cells to maintain themselves and to divide. To test our adaptation of the CPM we improve the realism of known simulations dealing with embryogenesis. Thanks to the CPM the shape of the cell is more complex than a simple pixel. Our simulation shows multilayer proliferation, similar to the bilayer proliferation described in the literature. In addition it reveals dynamic tissue regeneration, morphogenesis of disjunctive tissues and the dynamic stability of global shape.

International audience; By interacting with pathogens, lactic acid bacteria (LAB) are able to contribute to food safety. By means of their lactic acid production which induces pH decrease, LAB influence the growth of pathogens. The aim of this study is to model and simulate lactic acid production, pH evolution, according to carbohydrate concentration in media, temperature, water activity and ratio of both population.

In this article, we present a parallel image processing system based on the concept of reactive agents. Our system lies in the oRis language, which allows to describe finely and simply the agents’ behaviors to detect image features. We also present a method of segmentation using a multi-agent system, and two biological applications made with oRis. The stopping of this multi-agent system is implemented through a technique issued from immunology: the apoptosis.