Artículos y presentaciones en Congresos
URI permanente para esta colección
Examinar
Examinando Artículos y presentaciones en Congresos por Autor "Acosta, Gerardo G."
Mostrando 1 - 9 de 9
Resultados por página
Opciones de ordenación
- Artículo
Acceso Abierto ECOPAMPA: A new tool for automatic fish schools detection and assessment from echo data(2021) Villar, Sebastián; Madirolas, Adrián; Cabreira, Ariel; Rozenfeld, Alejandro F.; Acosta, Gerardo G.Accurate identification of aquatic organisms and their numerical abundance calculation using echo detection techniques remains a great challenge for marine researchers. A software architecture for echo data processing is presented in this article. Within it, it is discussed how to obtain energetic, morphometric and bathymetric fish school descriptors to accurately identify different fish-species. To accomplish this task it was necessary to have a development platform that allowed reading echo data from a particular echosounder, to detect fish aggregations and then to calculate fish school descriptors that would be used for fish-species identification, in an automatic way. This article also describes thoroughly the digital processing algorithms for this automatic detection and classification, as well as the automatic process required for surface and bottom line detection, which is necessary to determine the exploration range. These algorithms are implemented within the ECOPAMPA software, which is the first Argentinean system for marine species identification. Finally, a comparative result over experimental data of ECOPAMPA against EchoviewTM Software Pty Ltd (formerly Myriax Software Pty Ltd), is carefully examined. - Artículo
Acceso Abierto Efficient approach for OS-CFAR 2D technique using distributive histograms and breakdown point optimal concept applied to acoustic images(2019) Villar, Sebastián; Menna, Bruno V.; Torcida, Sebastián; Acosta, Gerardo G.In this work, a new approach to improve the algorithmic efficiency of the order statistic-constant false alarm rate (OSCFAR) applied in two dimensions (2D) is presented. OS-CFAR is widely used in radar technology for detecting moving objects as well as in sonar technology for the relevant areas of segmentation and multi-target detection on the seafloor. OSCFAR rank orders the samples obtained from a sliding window around a test cell to select a representative sample that is used to calculate an adaptive detection threshold maintaining a false alarm probability. Then, the test cell is evaluated to determine the presence or absence of a target based on the calculated threshold. The rank orders allow that OS-CFAR technique to be more robust to the presence of the speckle noise, but requires higher computational effort. This is the bottleneck of the technique. Consequently, the contribution of this work is to improve the OSCFAR 2D on-line computation with the distributive histograms and the optimal breakdown point optimal concept, mainly from the standpoint of efficient computation. The theoretical algorithm analysis is presented to demonstrate the improvement of this approach. Also, this novel efficient OS-CFAR 2D was contrasted experimentally on acoustic images. - Documento de conferencia
Acceso Abierto Ictiobot-40 a low cost AUV platform for acoustic imaging surveying(2019) Acosta, Gerardo G.; Mena, Bruno V.; Carlucho, Ignacio; De Paula, Mariano; Villar, Sebastián; Curti, Hugo J.; Rozenfeld, Alejandro F.; Vega, Roberto J. de la; Isasmendi, Agustín; Leegstra, Roberto C.; Arrien, Luis M.Autonomous Underwater Vehicles (AUVs) are suitable platforms for a wide type of applications in the oceanic environment. These applications are developed in various fields such as scientific surveying, off-shore industry and defense. The employment of AUVs requires less human support and reduces operation costs. Due to the changing marine environment these vehicles must deal with uncertain and hostile conditions to perform its tasks. In the marine robotics matter, the INTELYMEC group has developed in 2012 an AUV prototype called Ictiobot, a low cost experimental platform for multipurpose missions. In this paper an upgrade of the original prototype is presented, the Ictiobot-40, conceived to perform acoustic imaging surveying missions of up to two hours and maximum depths of 40 meters. The new software and hardware architectures and mechanical structure improvements, are detailed. In addition to these technical details, initial experimental results of the AUV performance in quiet waters will be discussed. Also, the new approaches for systems under development are presented. - Documento de conferencia
Acceso Abierto Integración de ROS y Tecnomatix para el desarrollo de gemelos digitales en sistemas de manufactura flexible(2020) Saavedra Sueldo, Carolina; Villar, Sebastián; De Paula, Mariano; Urrutia, Silvia B.; Acosta, Gerardo G.Los gemelos digitales emplean la simulación en conjunto con una variedad de datos provenientes de diferentes equipos y sistemas físicos de planta, para mantener actualizados continuamente sus modelos digitales del mundo en un esquema de retroalimentación en un entorno virtual que facilita la toma de decisiones. La heterogeneidad de hardware y software existente requiere del desarrollo de arquitecturas de software para que los diversos componentes se integren e interactúen mediante el intercambio de información. En este trabajo se propone el diseño y construcción de una arquitectura de software que integra un simulador de procesos de manufactura con el sistema operativo de robots (ROSRobot Operating System). La propuesta se prueba con el simulador Tecnomatix de Siemens® y la distribución libre ROS Melodic. Se presenta una instancia de la arquitectura de software para un caso de estudio complejo típico de plantas de manufactura y se demuestra su fácil integración con un sistema autónomo de toma de decisiones. - Artículo
Acceso Abierto Integration of ROS and Tecnomatix for the Development of Digital Twins Based Decision-Making Systems for Smart Factories(2021) Saavedra Sueldo, Carolina; Villar, Sebastián; De Paula, Mariano; Acosta, Gerardo G.Digital twins employs simulation in conjunction withvirtual environments and a variety of data coming from differentplant equipment and physical systems to continuously updatethe digital models of the world in a feedback loop schemeto facilitate the decision-making processes. The heterogeneityof existing hardware and software requires the developmentof software architectures able to deal with the informationexchange due to the integration and interaction of several systemcomponents and autonomous decision-making systems. In thiswork we propose the design and construction of a softwarearchitecture that integrates a manufacturing process simulatorwith the well-known robot operating system (ROS-RobotOperating System) to easily interchange information with anautonomous decision-making system. The proposal is tested withthe simulator Tecnomatix®and the freedistribution ROS Melodic.We present an instance of software architecture for a typicalcomplex case study of manufacturing plants and demonstrateits easy integration with an autonomous decision-making systembased on the reinforcement- learning paradigm. - Artículo
Acceso Abierto Navigation System for MACÁBOT an Autonomous Surface Vehicles Using GPS Aided Strapdown Inertial Navigation System(2019) Menna, Bruno V.; Villar, Sebastián; Acosta, Gerardo G.In this work the design, implementation and real-time tests of a navigation system for the autonomous surface vehicle MACÁBOT is presented. This vehiclerepresents a versatile platform to perform several tasks inthe marine environment, such as; ports maintenance,marine productive ecosystems studies and bathymetries.The navigation system is responsible for accuratelydetermining the position, velocity and attitude of the vehicle.It represents a fundamental component to autonomouslycarry out any of the aforementioned tasks. In this work, thenavigation system is developed based on a GPS aided strapdown inertial navigation system using an extended Kalmanfilter sensor fusion algorithm. In order to provide an adaptive approach to the sensor fusion algorithm tuning afuzzy inference system is used. The navigation system wasimplemented as a package for the Robot Operating System,benefiting from the advantages of heterogeneity, integrationand hardware abstraction. Real time tests of theMACÁBOT on a local creek were carried out, showingsatisfactory performance of the navigation system in bothposition and velocity estimates. In addition to these tests,simulations of GPS outages were carried out with theregistered data to evaluate the performance of thenavigation system in such cases. - Documento de conferencia
Acceso Abierto Optimización y control del flujo de materiales en procesos de producción flexibles utilizando aprendizaje profundo(2021) Saavedra Sueldo, Carolina; Perez Colo, Ivo; De Paula, Mariano; Villar, Sebastián; Acosta, Gerardo G.Industry 4.0, currently on the rise, demandsincreasing flexibility and adaptation of production systems tochanging products demands and external factors. The adaptationof the production systems implies frequent and often abruptchanges in the configurations of the shop floors and consequentlythe movement of materials must be re-planned. Materialhandlingis significant in terms of operative costs and times and it doesnot add value to the end products. It is desired to optimize theperformance of the system based onthe degree of movements,buffer usage and waiting times, such that the combinationof these minimizes the impact on the process costs. Machinelearning algorithms incombination with powerful computationalsimulators can be mutually leveraged to give rise to solvethese kinds of real-world problems, typical of smart factories.In this work, for the optimization approach, we develop aclosed-loop decision-making system with a deep reinforcementlearning algorithm based on a discrete-event simulation modelfor material handling. In addition, our proposed approach usesthe communication architecture Simulai, which allows interfacinga computational discrete-event simulator and the proposed deep learning-based decision-making algorithm. The functionality ofour proposal is evidenced through the obtained results and anoptimal solution for the problem stated is reached, proving thatan intelligent agentcan collaborate in making multiple decisionsfor smart factories. - Artículo
Acceso Abierto Particle Filter based Autonomous Underwater Vehicle Navigation System aided thru acoustic communication ranging(2020) Menna, Bruno V.; Villar, Sebastián; Acosta, Gerardo G.Autonomous Underwater Vehicles (AUVs) are platforms suitable for a wide variety of applications in the marine environment with economic and operational advantages. In these applications an AUV performs a given task as a mission. During the mission execution, the AUV will move around the environment following paths that allow it to fulfill the mission's objectives. To achieve this, a reliable Navigation System (NS) is required. In addition to this, the current operating concept includes the deployment of multiple AUVs on a given area, thus a communication system between vehicles is also required. In the underwater environment both navigation and communication systems deals with the particular characteristics of the medium that limits the use of conventional techniques. In this work, a complete NS for an AUV is presented. The developed NS is based on an inertial navigation scheme with velocity and position aiding. The position aiding takes advantage of the communication system onboard the vehicle, which avoids the use of additional positioning systems. The fundamentals of the applied solutions are described and experimental results and implementation details are provided. Also conclusions and future works are presented. - Documento de conferencia
Acceso Abierto Sistema inteligente para la detección de fallas basado en redes profundas auto-ajustables(2022) Pérez Colo, Ivo; Saavedra Sueldo, Carolina; De Paula, Mariano; Roark, Geraldina; Villar, Sebastián; Acosta, Gerardo G.La creciente complejidad de los sistemas industriales ha fomentado el surgimiento de nuevas técnicas de análisis de datos para apoyar a los procesos de toma de decisiones. Concretamente, los modelos basados en redes neuronales profundas constituyen una alternativa promisoria para diversas aplicaciones de detección, clasificación y predicción de defectos o fallas que abarca aplicaciones desde el control de calidad de los productos, identificación de defectos en los procesos en una línea de producción hasta predicción de fallas de los equipos tecnológicos. Sin embargo, el éxito de dichos modelos depende sensiblemente de la elección de sus hiper-parámetros para lo cual se requiere de un exhaustivo proceso de configuración que, hoy en día, demanda un alto grado de conocimiento experto. En este contexto, el presente trabajo propone un Sistema Inteligente basado en redes neuronales profundas, dotado con un sistema de auto-ajuste de sus hiper-parámetros, para la detección de defectos y fallas. Dicho sistema integra un algoritmo de Optimización Bayesiana para encontrar la combinación óptima de los hiperparámetros que permita alcanzar el mejor desempeño posible del sistema. El sistema inteligente propuesto se prueba en dos casos de estudio de diferente naturaleza y los resultados alcanzado demuestran la efectividad de la propuesta.