Publication: Arquitectura para la gestión y coordinación de sistemas multisensor
Loading...
Identifiers
Publication date
2015-06
Defense date
2015-06-08
Authors
Tutors
Journal Title
Journal ISSN
Volume Title
Publisher
Impact
Export
Abstract
Hoy en día podemos encontrarnos con numerosos entornos que cuentan con un gran
conjunto de sensores heterogéneos espacialmente distribuidos. Entornos que pueden ir desde sistemas de vigilancia que incluyen un despliegue de múltiples sensores como cámaras o sensores de localización, a entornos de monitorización de procesos industriales, donde se supervisa cada uno de los procesos para comprobar el correcto funcionamiento del sistema. Inclusive actualmente podemos hablar de términos más modernos como el de Smart Cities que permiten la optimización de los recursos de una ciudad a través de su constante monitorización. En todos estos entornos es común encontrarnos con multitud de sensores percibiendo información del entorno.
Estos entornos son de indudable utilidad en cada uno de sus ámbitos, y según avanza la tecnología podemos mejorarlos con la integración de más y mejores sensores. Esto resulta de utilidad porque nos permite monitorizar con más precisión y calidad las diferentes entidades o parámetros del entorno. Sin embargo esto puede suponer un problema a la hora de gestionar y coordinar diferentes tareas sobre el entorno, debido principalmente a la gran cantidad de información generada que tiene que ser analizada. En este sentido, la fusión de datos se volverá un aliado indispensable para mejorar los procesos de análisis y toma de decisiones. En estos entornos además podemos encontrarnos con sensores u actuadores que pueden
requerir de cierta gestión, como por ejemplo el manejo de una cámara PTZ para monitorizar regiones o entidades de interés del entorno. Esta gestión se puede hacer de manera manual a través de un operador humano, pero puede convertirse en una tarea compleja cuando contamos con multitud de sensores similares que puedan requerir una gestión coordinada. En este sentido, en la tesis se abordará este problema de gestión y coordinación desde el punto de vista de un sistema multi-agente. Para ello diseñaremos una arquitectura general, que podrá ser aplicada a diferentes casos de uso. Sobre este diseño, evaluaremos su aplicación a dos entornos diferentes. El primer entorno consistirá en un sistema de vigilancia multi-cámara,
donde será necesario realizar un control autónomo de las cámaras PTZ para monitorizar las diferentes entidades de interés. El segundo entorno se desarrollará sobre un sistema de vigilancia marítima, donde además de las cámaras podremos contar con otro tipo de sensores como radares o estaciones AIS. La experimentación llevada a cabo en esta tesis a través de la aplicación multi-cámara, ha permitido evaluar la arquitectura desde el punto de vista de la integración en un entornoreal. La aplicación sobre el entorno marítimo, en este caso desarrollado sobre un entorno
simulado, nos ha permitido además evaluar la adecuación de de la arquitectura a diferentes sensores e información de las entidades. En ambos casos hemos podido observar la utilidad del sistema multi-agente desarrollado, así como su adecuación a este tipo de entornos, que dada su naturaleza son inherentemente distribuidos.
Nowadays it is common to find several heterogeneous sensors spatially distributed in environments. Environments that can range from surveillance systems that include a set of multiple sensors such as cameras or location sensors. To environments monitoring industrial processes, where it is important to monitor several processes to ensure a proper system operation. Moreover we can talk about more modern terms as Smart Cities, which allow optimizing the city resources through its constant monitoring. Thus, in all those deployments, it is common to find multiple sensors perceiving information from the environment. These environments are undoubtedly useful in each of their fields, and as technology advances, we can improve the integration of more and better sensors. This is useful because it allows a better and more accurately monitoring of the different entities or environmental parameters. However this can be a problem when it comes to managing and coordinating different tasks over the environment, mainly due to the large amount of information generated required to be analyzed. In this sense, data fusion will become an indispensable ally to improve the analysis and decision making processes. Moreover, in these environments we can find multiple sensors or actuators that may require some kind of management, such as managing a PTZ camera to monitor regions or entities of interest. This management can be done manually by a human operator, but can become a complex task when we have multiple sensors requiring a coordinated management. In this sense, this thesis tackle the problem of multi-sensor management and coordination from the point of view of a multi-agent system. It is proposed a general multi-agent design that can be applied to different use cases. This design will be evaluated in two different environments. The first one will consist on a multi-camera surveillance system, where the multi-agent system must achieve an autonomous control of PTZ cameras to monitor different entities of interest. The second environment will rely on a maritime surveillance system, where in addition to the cameras, we will introduce other sensor types such as radar or AIS stations. Some experiments were carried out to evaluate the multi-agent system designed. The experiments done in the multi-camera application, allowed us to evaluate the architecture from the point of view of its integration in a real environment. Meanwhile the application in the maritime environment, developed over a simulated environment, let us evaluate the suitability of the architecture to different sensors and information. In both cases we have been able to observe the usefulness of the developed multi-agent system and its adaptation to these environments inherently distributed.
Nowadays it is common to find several heterogeneous sensors spatially distributed in environments. Environments that can range from surveillance systems that include a set of multiple sensors such as cameras or location sensors. To environments monitoring industrial processes, where it is important to monitor several processes to ensure a proper system operation. Moreover we can talk about more modern terms as Smart Cities, which allow optimizing the city resources through its constant monitoring. Thus, in all those deployments, it is common to find multiple sensors perceiving information from the environment. These environments are undoubtedly useful in each of their fields, and as technology advances, we can improve the integration of more and better sensors. This is useful because it allows a better and more accurately monitoring of the different entities or environmental parameters. However this can be a problem when it comes to managing and coordinating different tasks over the environment, mainly due to the large amount of information generated required to be analyzed. In this sense, data fusion will become an indispensable ally to improve the analysis and decision making processes. Moreover, in these environments we can find multiple sensors or actuators that may require some kind of management, such as managing a PTZ camera to monitor regions or entities of interest. This management can be done manually by a human operator, but can become a complex task when we have multiple sensors requiring a coordinated management. In this sense, this thesis tackle the problem of multi-sensor management and coordination from the point of view of a multi-agent system. It is proposed a general multi-agent design that can be applied to different use cases. This design will be evaluated in two different environments. The first one will consist on a multi-camera surveillance system, where the multi-agent system must achieve an autonomous control of PTZ cameras to monitor different entities of interest. The second environment will rely on a maritime surveillance system, where in addition to the cameras, we will introduce other sensor types such as radar or AIS stations. Some experiments were carried out to evaluate the multi-agent system designed. The experiments done in the multi-camera application, allowed us to evaluate the architecture from the point of view of its integration in a real environment. Meanwhile the application in the maritime environment, developed over a simulated environment, let us evaluate the suitability of the architecture to different sensors and information. In both cases we have been able to observe the usefulness of the developed multi-agent system and its adaptation to these environments inherently distributed.
Description
Keywords
Inteligencia artificial, Control automático, Multisensor, Vigilancia, Modelos de fusión, Arquitectura multi-agente, Cámaras PTZ