Publication: Desarrollo de un algoritmo de auto-calibración de sistemas estéreo basado en el uso del UV-Disparity : aplicación a la odometría visual
Loading...
Identifiers
Publication date
2012-10
Defense date
2012-10-26
Authors
Advisors
Tutors
Journal Title
Journal ISSN
Volume Title
Publisher
Impact
Export
Abstract
El objetivo de este proyecto es el de presentar una solución que calibre de forma autónoma los parámetros extrínsecos de un par de cámaras estéreo cuyo fin es trazar la trayectoria del vehículo mediante un algoritmo de odometría visual. La calibración de la cámara ha de realizarse en cada instante de captura de la cámara para así poder minimizar los errores debidos a posibles vibraciones del vehículo mientras este está en movimiento. Gracias a ello, los resultados serán más precisos y robustos. Para ello, se proceder á a introducir las fórmulas que relacionan el entorno con el sistema de la cámara, y poder con ello efectuar una calibración de ellas antes de calcular la trayectoria del vehículo. El algoritmo ha sido diseñado en CUDA (Compute Unified Device Architecture) para poder satisfacer con la demanda de bajo tiempo de cómputo necesaria para poder llevar a cabo la empresa. El código implementado será introducido en el prototipo de vehículo inteligente de la UC3M (Universidad Carlos III de Madrid), ivvi (Intelligent Vehicle Based On Visual Information) 2.0 para poder formar un sistema de navegación dependiente únicamente de los sensores instalados en él. ______________________________________________________________________________________________________________
The aim of this project is to present a solution that calibrates automatically the extrinsic parameters of stereo vision pairs of cameras, whose objective is to determine the path of the vehicle due to a visual odometry algorithm. The calibration of the camera must be done in each frame, in order to minimize the errors that occur because of the car's vibrations while it is being driven. Thanks to this improvements, the results will be more accurate and robust. For this purpose, formulas that relate the environment with the camera coordinate system will be introduced, so that calibration takes place before the algortihm calculates the current position of the vehicle. The algorithm has been designed in CUDA (Compute Unifi ed Device Architecture), in order to satisfy with the low computation time request that is needed in this problem. The implemented code will be used in the intelligent vehicle owned by the UC3M (Universidad Carlos III de Madrid), called ivvi (Intelligent Vehicle Based On Visual Information) 2.0, to create a navigation system that only depends on the on-board sensors.
The aim of this project is to present a solution that calibrates automatically the extrinsic parameters of stereo vision pairs of cameras, whose objective is to determine the path of the vehicle due to a visual odometry algorithm. The calibration of the camera must be done in each frame, in order to minimize the errors that occur because of the car's vibrations while it is being driven. Thanks to this improvements, the results will be more accurate and robust. For this purpose, formulas that relate the environment with the camera coordinate system will be introduced, so that calibration takes place before the algortihm calculates the current position of the vehicle. The algorithm has been designed in CUDA (Compute Unifi ed Device Architecture), in order to satisfy with the low computation time request that is needed in this problem. The implemented code will be used in the intelligent vehicle owned by the UC3M (Universidad Carlos III de Madrid), called ivvi (Intelligent Vehicle Based On Visual Information) 2.0, to create a navigation system that only depends on the on-board sensors.
Description
Keywords
Tecnología automovilística, Odometrí a visual, Cámaras estéreo, Algoritmos