Deep space missions typically face navigation challenges for vision-based algorithms, which need to be tested with meaningful data. Several techniques can be employed, each with a different level of abstraction: the higher the abstraction, the less representative the generated data. To face these issues, Hardware-in-the-Loop tests have gained popularity thanks to their ability to provide high-fidelity simulations. Using a mockup of the real target and a camera mounted on high torque robotic manipulators, it is possible to simulate a real life scenario. These tests are extremely important for missions where in situ tests are prohibitive in terms of cost or simply not possible. However, these datasets must be acquired with a good level of consistency: both in terms of preservation of the trajectory constraints, and from a dynamic viewpoint. The thesis aims to produce a mapping technique which can be applied to the RAFFAELLO testbed: a procedure that maps relative orbital dynamics into the laboratory facility, ensuring both high-fidelity and dynamic-consistent results. Robotics fundamentals are combined with nonlinear optimization, where the mapping solution is solved in a constrained environment. Digital twin methodologies are used to model the hardware of the facility, in order to verify and validate the optimization results, while Hardware-in-the-Loop tests are carried out to acquire datasets in the existing facility.
Le missioni nello spazio profondo affrontano tipicamente sfide di navigazione per algoritmi vision-based, che devono essere testati con dati significativi. Diverse tecniche possono essere impiegate, ciascuna con un diverso livello di astrazione: più alto è il livello di astrazione, meno rappresentativi sono i dati generati. Per affrontare questi problemi, i test Hardware-in-the-Loop hanno guadagnato popolarità grazie alla loro capacità di fornire simulazioni ad alto livello di fedeltà. Utilizzando un mockup del target reale e una fotocamera montata su manipolatori robotici high torque, è possibile simulare uno scenario reale. Questi test sono estremamente importanti per le missioni in cui i test in situ sono proibitivi in termini di costi o semplicemente non possibili. Tuttavia, questi dataset devono essere acquisiti con un buon livello di consistenza: sia in termini di preservazione dei vincoli di traiettoria, sia dal punto di vista dinamico. La tesi mira a produrre una tecnica di mapping che possa essere applicata al testbed RAFFAELLO: una procedura che mappa le dinamiche relative in laboratorio, garantendo sia risultati high-fidelity, sia consistenti dal punto di vista dinamico. I fondamenti della robotica sono combinati con nonlinear optimization, dove la soluzione di mapping è risolta in un ambiente vincolato. Le metodologie di digital twin sono utilizzate per modellare l'hardware della facility, al fine di verificare e validare i risultati dell'ottimizzazione, mentre i test Hardware-in-the-Loop vengono eseguiti per acquisire dataset nella facility esistente.
Dynamics-consistent trajectory mapping and emulation in the RAFFAELLO robotic testbed: methods with digital twin validation and dataset generation
Belletti, Stefano
2024/2025
Abstract
Deep space missions typically face navigation challenges for vision-based algorithms, which need to be tested with meaningful data. Several techniques can be employed, each with a different level of abstraction: the higher the abstraction, the less representative the generated data. To face these issues, Hardware-in-the-Loop tests have gained popularity thanks to their ability to provide high-fidelity simulations. Using a mockup of the real target and a camera mounted on high torque robotic manipulators, it is possible to simulate a real life scenario. These tests are extremely important for missions where in situ tests are prohibitive in terms of cost or simply not possible. However, these datasets must be acquired with a good level of consistency: both in terms of preservation of the trajectory constraints, and from a dynamic viewpoint. The thesis aims to produce a mapping technique which can be applied to the RAFFAELLO testbed: a procedure that maps relative orbital dynamics into the laboratory facility, ensuring both high-fidelity and dynamic-consistent results. Robotics fundamentals are combined with nonlinear optimization, where the mapping solution is solved in a constrained environment. Digital twin methodologies are used to model the hardware of the facility, in order to verify and validate the optimization results, while Hardware-in-the-Loop tests are carried out to acquire datasets in the existing facility.| File | Dimensione | Formato | |
|---|---|---|---|
|
2026_03_Belletti.pdf
accessibile in internet per tutti
Dimensione
18.31 MB
Formato
Adobe PDF
|
18.31 MB | Adobe PDF | Visualizza/Apri |
|
2026_03_Belletti_Executive_Summary.pdf
accessibile in internet per tutti
Dimensione
2.18 MB
Formato
Adobe PDF
|
2.18 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/252988