In manufacturing, robots often navigate predetermined paths manually programmed via a teach pendant, particularly in tasks such as welding and gluing. While this method is accessible and doesn’t require advanced programming skills, its repetitive nature throughout the day becomes time-consuming and monotonous. The implementation of an automated system capable of generating robot paths directly from images or CAD models presents itself as an ideal solution, offering notable enhancements in efficiency and usability. This thesis is dedicated to the development of an alternative approach to robot calibration, focusing specifically on the intricate process of mapping points from camera images to real-world coordinates. Previous methodologies, which relied on trial and error to determine the optimal homogeneous transformation, have proven inefficient, especially when relocating the robot. This thesis proposes an innovative algorithm inspired by the Iterative Closest Point (ICP) methodology. This approach begins by aligning points from the image to real-world coordinates to address rotation discrepancies, followed by the application of optimization techniques to fine-tune translation parameters. The result is a calibration process that is not only more robust but also adaptable to various scenarios.
Nell’ambito della produzione, i robot spesso seguono percorsi predeterminati programmati manualmente tramite un teach pendant, specialmente in compiti come la saldatura e l’incollaggio. Nonostante questa procedura è facilmente accessibile e non richiede competenze avanzate di programmazione, la sua ripetitività diventa onerosa in termini di tempo e monotona nel corso della giornata. L’implementazione di un sistema automatizzato in grado di generare percorsi robotici direttamente da immagini o modelli CAD, si presenta come una soluzione ideale, offrendo notevoli miglioramenti in termini di efficienza. Questa tesi è dedicata allo sviluppo di un approccio alternativo alla calibrazione dei robot, focalizzandosi nello specifico sul processo di mappatura dei punti dalle immagini della telecamera alle coordinate del mondo reale. In progetti precedenti nello stesso contesto, che si basavano su provare a tentativi per determinare la trasformazione omogenea ottimale, si sono dimostrate inefficienti, specialmente in scenari che coinvolgevano lo spostamento del robot in un’altra posizione. Questa tesi propone un algoritmo alternativo ispirato alla metodologia del Iterative Closest Point (ICP), il quale si basa sull’allineando dei punti dall’immagine alle coordinate del mondo reale per correggere le discrepanze di rotazione, seguito da tecniche di ottimizzazione per migliorare i parametri di traslazione. Il risultato ottenuto è un processo di calibrazione non solo più robusto, ma anche adattabile a vari contesti.
Precise Calibration for Robot Path Planning on World Plane
DUAN, ZEBIN
2023/2024
Abstract
In manufacturing, robots often navigate predetermined paths manually programmed via a teach pendant, particularly in tasks such as welding and gluing. While this method is accessible and doesn’t require advanced programming skills, its repetitive nature throughout the day becomes time-consuming and monotonous. The implementation of an automated system capable of generating robot paths directly from images or CAD models presents itself as an ideal solution, offering notable enhancements in efficiency and usability. This thesis is dedicated to the development of an alternative approach to robot calibration, focusing specifically on the intricate process of mapping points from camera images to real-world coordinates. Previous methodologies, which relied on trial and error to determine the optimal homogeneous transformation, have proven inefficient, especially when relocating the robot. This thesis proposes an innovative algorithm inspired by the Iterative Closest Point (ICP) methodology. This approach begins by aligning points from the image to real-world coordinates to address rotation discrepancies, followed by the application of optimization techniques to fine-tune translation parameters. The result is a calibration process that is not only more robust but also adaptable to various scenarios.File | Dimensione | Formato | |
---|---|---|---|
Duan_Zebin.pdf
accesso riservato
Dimensione
33.74 MB
Formato
Adobe PDF
|
33.74 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/62370