Riassunto analitico
Autonomous driving is a topic of particular interest for automotive companies nowadays.
It will assume a major role in mobility indeed influencing the world's car industry.
According to experts, moving towards such a direction will enhance the production and the development of automotive oriented sensors.
Stereo cameras are a family of sensors belonging to such area of interest.
The ability of recognizing objects, reconstruct the surrounding environment in 3D, make them suitable for self-driving cars, in order to avoid obstacles or to make the right decision autonomously.
Considering that the future of mobility is going in this direction, of relevant importance is the study, development and comparison of these different sensors.
Standards for the measure of performance of stereo cameras have been proposed in the recent few years, giving the possibility of comparison among different instruments.
The area of interest however has always been the close range (0-2m), this because of the usage of stereo cameras, was almost in the total of cases, in indoor environment or used for image processing, where short distances are used.
With the increase of 3D sensors production and their suitability for autonomous driving, the focus has been moved to the outdoor scenario.
The proposed work is focused on the development of a standard and repeatable methodology, to determine the performances of a stereo camera in a wider range (0-20m), this indeed will also be helpful to compare different sensors.
This work of thesis aims to the realization of novel measurement procedure for the characterization
and comparison of stereo vision systems. In particular, the activities have been focused on
range error, warm-up time, stability, RMS error and fill rate determination.
|
Abstract
Autonomous driving is a topic of particular interest for automotive companies nowadays.
It will assume a major role in mobility indeed influencing the world's car industry.
According to experts, moving towards such a direction will enhance the production and the development of automotive oriented sensors.
Stereo cameras are a family of sensors belonging to such area of interest.
The ability of recognizing objects, reconstruct the surrounding environment in 3D, make them suitable for self-driving cars, in order to avoid obstacles or to make the right decision autonomously.
Considering that the future of mobility is going in this direction, of relevant importance is the study, development and comparison of these different sensors.
Standards for the measure of performance of stereo cameras have been proposed in the recent few years, giving the possibility of comparison among different instruments.
The area of interest however has always been the close range (0-2m), this because of the usage of stereo cameras, was almost in the total of cases, in indoor environment or used for image processing, where short distances are used.
With the increase of 3D sensors production and their suitability for autonomous driving, the focus has been moved to the outdoor scenario.
The proposed work is focused on the development of a standard and repeatable methodology, to determine the performances of a stereo camera in a wider range (0-20m), this indeed will also be helpful to compare different sensors.
This work of thesis aims to the realization of novel measurement procedure for the characterization
and comparison of stereo vision systems. In particular, the activities have been focused on
range error, warm-up time, stability, RMS error and fill rate determination.
|
