Publication: Control de sensores electro-ópticos en aviones no tripulados y tratamiento de imágenes utilizando máquinas de vectores soporte
Loading...
Identifiers
Publication date
2010-06
Defense date
2010-06-10
Authors
Tutors
Journal Title
Journal ISSN
Volume Title
Publisher
Impact
Export
Abstract
Hoy en día, la revolución tecnológica, permite realizar cosas que hasta hace unos años eran impensables; como por ejemplo, identificar personas por los rasgos faciales o compartir gran cantidad de información a través de una red llamada internet. El sector militar aeronáutico es uno de los que más invierte en investigación, desarrollo e implantación de los más nuevos y sofisticados avances tecnológicos y que en el futuro se aplicarán en el sector civil. Actualmente, el boom aeronáutico pasa por el desarrollo de aviones no tripulados, los llamados UAVs (del inglés, Unmanned Aerial Vehicles). Las empresas punteras del sector, como EADS (European Aeronautics Defence and Space Company) o Boeing, están invirtiendo en el desarrollo de UAVs cada vez más sofisticados; capaces de realizar misiones de mayor duración y con la tecnología más puntera, con el objetivo de llevar a cabo las misiones requeridas con total éxito. Este proyecto se ha planteado dentro de la empresa EADS, gracias a los nuevos desarrollos que se están llevando a cabo en la división de Defence & Security. Este proyecto, surgió a raíz del nuevo desarrollo llamado UAV ATLANTE (Avión Táctico de Largo Alcance No Tripulado) el primer UAV español, el cual está desarrollando para el ejército español la empresa EADS. Entre las misiones más importantes que desarrollan los UAVs se encuentran las misiones de inteligencia, vigilancia, adquisición de blancos y reconocimiento. Para ello, el avión lleva incorporado en la parte inferior del fuselaje un sensor EO (Electro-Óptico), para el cual se va a desarrollar este proyecto. A pesar de que el proyecto surgiera a raíz del desarrollo del UAV ATLANTE, éste es aplicable a cualquier UAV que esté provisto de un sensor EO, en cuyos computadores o en los de la estación base, sea posible integrar los algoritmos desarrollados. El proyecto se ha subdividido en dos subproyectos de disciplinas muy dispares, que han sido aplicados de forma combinada para lograr un fin común. El primer subproyecto, desarrollado en la empresa EADS, consiste en la implementación de los algoritmos de geo-apuntamiento (geopointing) y geo-localización (geolocation). El algoritmo de geo-apuntamiento permite calcular los ángulos del sensor EO para que éste apunte a una latitud, longitud y altura determinadas por el operador del UAV situado en tierra. El algoritmo de geo- localización permite calcular la latitud, longitud y altura de la posición geográfica a la que la cámara apunte en el momento que el operador le indique al UAV. Una vez implementados los dos algoritmos, se realizará una batería de pruebas para verificar el correcto funcionamiento de los mismos. Las pruebas se realizarán en primer lugar en un PC, así se eliminarán posibles fallos de codificación, y a continuación, se realizará una integración de los algoritmos en un entorno de simulación. Esta integración se llevará a cabo en un banco de pruebas perteneciente a EADS. Las pruebas de integración en banco permitirán comprobar el correcto funcionamiento de los algoritmos de manera visual. La cámara del sensor EO del UAV es capaz de capturar imágenes en tiempo real. El procesado y análisis de estas imágenes para la detección de posibles objetivos como, fuegos, personas, con- voyes de misión humanitaria. . . son fundamentales para el correcto desarrollo de las misiones de un UAV. En la segunda parte del proyecto, la cual ha sido desarrollada en la UC3M (Universidad Carlos III de Madrid), se quiere realizar una implementación de un clasificador de imágenes, y discutir en torno a la viabilidad de embarcar la implementación en el UAV o ejecutarla en la estación de tierra. Para su realización se ha optado por utilizar una implementación de las máquinas de vectores soporte (SVM, Support Vector Machines); ya que permiten realizar una clasificación máquina con un alto porcentaje de aciertos, a la vez que se minimiza la cantidad de información necesaria. _______________________________________________________________________________________________
Nowadays, technological revolution has allowed to develop features absolutely unbelievable some years ago. For example, identifying people by facial features or sharing a wealth of information through a network called the Internet. The military aircraft sector is one of the largest investors in research, development and implementation on the newest and most sophisticated technological advances which will be applied in the civilian sector on the future. Currently, the aviation boom goes through the development of UAVs (Unmanned Aerial Vehicles). Leader companies, such as EADS (European Aeronautics Defence and Space Company) and Boeing, are investing in the development of increasingly sophisticated UAVs, capable to perform longer missions with the leading technology, with the aim to obtain resounding successes on the required missions. This project has arisen within the company EADS, thanks to new developments that are being developed in the division of Defence & Security. This project arose, because of the new development called UAV ATLANTE (Spanish long range tactical aircraft unmanned) the first Spanish UAV, which is developed for the Spanish army by EADS. Among the most important missions that are being developed by UAVs are the missions of intelligence, surveillance, target acquisition and recognition. For this, the aircraft carries incorporated in the belly a EO (Electro-Optical) sensor, for which it will develop this project. In spite of the fact that the project arose because of the development of the UAV ATLANTE, this is applicable to any UAV that be provided of a EO sensor, in which whether on-board computers or those of the base station, could be possible to integrate in the developed algorithms. The project has been subdivided in two very different subproyects that have been applied in combination to achieve a common goal. The first subproject developed at EADS is the implementation of the algorithms for geopoin- ting and geolocation. The geopointing algorithm works out the angles of the EO sensor pointing to the latitude, longitude and height determined by the UAV operator located on land. The geolocation algorithm works out the latitude, longitude and height of the geographic position where the camera is focusing at the moment that the user indicates to the UAV. Once implemented the two algorithms, a battery of tests will be performed to verify proper operation thereof. The tests will be carry out first on a PC, to remove potential coding errors. Then there will be an integration of algorithms in a simulation environment. This integration will be carry out in a test bench belonging to EADS. The integration test bench will allow testing the proper functioning of the algorithms in a visual way. The camera of the EO sensor of the UAV is capable of capturing images in real time. Processing and analysing these images to detect possible targets such as fires, people who need rescuing, convoys of humanitarian tasks,... are essential for the correct development of the missions of an UAV. In the second part of the project, which has been developed in the UC3M (Carlos III University of Madrid), an implementation of an image classifier has been done, and also discussion about the feasibility of deployment on board the UAV or run in the ground station. To achieve this target, an implementation of Support Vector Machines (SVMs) has been chosen, as they allow a classification machine with a high percentage of hits, while minimizing the amount of necessary information.
Nowadays, technological revolution has allowed to develop features absolutely unbelievable some years ago. For example, identifying people by facial features or sharing a wealth of information through a network called the Internet. The military aircraft sector is one of the largest investors in research, development and implementation on the newest and most sophisticated technological advances which will be applied in the civilian sector on the future. Currently, the aviation boom goes through the development of UAVs (Unmanned Aerial Vehicles). Leader companies, such as EADS (European Aeronautics Defence and Space Company) and Boeing, are investing in the development of increasingly sophisticated UAVs, capable to perform longer missions with the leading technology, with the aim to obtain resounding successes on the required missions. This project has arisen within the company EADS, thanks to new developments that are being developed in the division of Defence & Security. This project arose, because of the new development called UAV ATLANTE (Spanish long range tactical aircraft unmanned) the first Spanish UAV, which is developed for the Spanish army by EADS. Among the most important missions that are being developed by UAVs are the missions of intelligence, surveillance, target acquisition and recognition. For this, the aircraft carries incorporated in the belly a EO (Electro-Optical) sensor, for which it will develop this project. In spite of the fact that the project arose because of the development of the UAV ATLANTE, this is applicable to any UAV that be provided of a EO sensor, in which whether on-board computers or those of the base station, could be possible to integrate in the developed algorithms. The project has been subdivided in two very different subproyects that have been applied in combination to achieve a common goal. The first subproject developed at EADS is the implementation of the algorithms for geopoin- ting and geolocation. The geopointing algorithm works out the angles of the EO sensor pointing to the latitude, longitude and height determined by the UAV operator located on land. The geolocation algorithm works out the latitude, longitude and height of the geographic position where the camera is focusing at the moment that the user indicates to the UAV. Once implemented the two algorithms, a battery of tests will be performed to verify proper operation thereof. The tests will be carry out first on a PC, to remove potential coding errors. Then there will be an integration of algorithms in a simulation environment. This integration will be carry out in a test bench belonging to EADS. The integration test bench will allow testing the proper functioning of the algorithms in a visual way. The camera of the EO sensor of the UAV is capable of capturing images in real time. Processing and analysing these images to detect possible targets such as fires, people who need rescuing, convoys of humanitarian tasks,... are essential for the correct development of the missions of an UAV. In the second part of the project, which has been developed in the UC3M (Carlos III University of Madrid), an implementation of an image classifier has been done, and also discussion about the feasibility of deployment on board the UAV or run in the ground station. To achieve this target, an implementation of Support Vector Machines (SVMs) has been chosen, as they allow a classification machine with a high percentage of hits, while minimizing the amount of necessary information.
Description
Keywords
Sensores electro-ópticos, Optoelectrónica, Aeronáutica, Algortimos