| KEYWORD |
Vision-based pose estimation of an unknown and uncooperative object for on-orbit servicing and debris removal applications
Tesi esterna in azienda
Parole chiave AUTONOMOUS NAVIGATION, EMBEDDED SYSTEMS, SATELLITE NAVIGATION, VISION
Riferimenti MARTINA MAMMARELLA
Riferimenti esterni Fabrizio Dabbene
Gruppi di ricerca 04-Automazione e Robotica
Tipo tesi SPERIMENTAL AND MODELLING
Descrizione Nowadays all the major space agencies and industries are undertaking the enhancement of the automation capabilities of spacecraft. Autonomous operations are extremely advantageous for space missions, with a wide range of possible applications. For example, autonomous relative navigation around unknown and uncooperative objects is particularly appealing. Precise pose and motion estimation of an uncooperative object, such as a Resident Space Object (RSO) has a potential utilization in the domain of space debris removal and on-orbit servicing.
This thesis is aimed at developing a simulation framework where the potentialities of vision-based pose estimation, combined with filtering methods, are investigated. Vision Based systems represent nowadays a promising tool to obtain satellite relative trajectory estimation within an optimal level of accuracy. In this way, the camera equipped on board the satellite is not only a support system to investigate the RSO but also to extend the satellite sensors compound for the autonomous navigation task. Indeed, the first pose estimate retrieved by the camera, in terms of relative position and attitude between the observing satellite and the RSO, can be fed to the navigation filter and fused with the other signals provided by e.g. GPS, Sun sensors, Star trackers, IMU, etc.
The activity aims to train new professionals in the areas of autonomous systems, and more specifically on vision-based navigation. The acquired competencies can be also applied to a wider plethora of applications, ranging from precision farming to autonomous driving and, more in general, to the Industry 4.0 framework.
The candidate will be in charge of developing a vision-based navigation system for a Low Earth Orbit RSO, which represents a debris that shall be captured and driven towards a re-entry trajectory. The candidate will be involved from the modelling of the sensors equipped on board (including the vision system) to the design of (robust) filtering technique for estimating the pose of the RSO along the orbit. The candidate will investigate different filtering techniques and will be in charge of selecting the one that best fit with the specific mission scenario, computational limitations, and robustness requirements.
Vedi anche cnr - thesis proposal on vision-based navigation for debris removal.pdf
Conoscenze richieste • Estimation theory sensor fusion, computer vision, robotics
• MATLAB/Simulink and/or Python/C++ coding skills
• spoken/written English (excellent)
Scadenza validita proposta 31/12/2023
PROPONI LA TUA CANDIDATURA