This thesis presents the design, implementation and validation of an integrated robotic system for the automated inspection and surface finishing of complex surfaces, based on the combination of a line laser scanner, an RGB-D camera and two collaborative robots. The goal was to develop a complete and autonomous workflow able to reconstruct the 3D geometry of a component, detect surface defects using robust algorithms and perform automatic correction of imperfections through robotic finishing operations with force feedback control. The proposed system integrates high-resolution optical profilometry with robotic manipulation, enabling accurate acquisition of the surface topography and subsequent correction of defects without human intervention. The initial reconstruction of the component, performed using the RGB-D camera, allows the areas of interest to be selected, which are then analyzed in greater detail by the Gocator 2629 laser profilometer. The defect identification is based on a mobile windows RANSAC interpolation algorithm, capable of robustly isolating local geometric anomalies even in the presence of noise. The detected defects are then grouped and transformed into operational trajectories for robotic sanding, validated in the RoboDK simulation environment and executed via realtime force control. The system calibration between the two Techman TM12 robots ensured coherent transformation of reference frames, achieving an average positional error of approximately 3 mm, fully compatible with the tolerances required for automated finishing. The results confirmed the system’s ability to detect real and artificial defects with high spatial accuracy and to maintain stable contact forces during finishing, demonstrating consistency between simulated and real trajectories. Overall, the developed system demonstrates the feasibility of a complete automatized robotic cell for inspection and surface finishing, able to reduce the human intervention while guaranteeing at the same time precision, repeatability and efficiency. The thesis work represents a concrete contribution towards adaptive and intelligent robotic systems for the identification and correction of defects, aligned with the principles of Industry 4.0.
Il presente lavoro di tesi riguarda la progettazione, l’implementazione e lavalidazione di un sistema robotico integrato per l’ispezione automatica e la finitura superficiale di superfici complesse, combinando un profilometro laser, una telecamera RGB-D e robot collaborativi. L’obiettivo è stato lo sviluppo di un processo completo e autonomo capace di ricostruire la geometria 3D di un componente, riconoscere i difetti superficiali attraverso un algoritmo robusto ed eseguire un’operazione di finitura superficiale utilizzando un’operazione robotica con il controllo di forza. Il sistema proposto integra la profilometria ad alta risoluzione con la manipolazione robotica, permettendo acquisizioni accurate della topografia della superficie e la correzione dei difetti senza l’intervento umano. La ricostruzione del componente, svolta tramite una telecamera RGB-D, permette la selezione della regione di interesse, che viene successivamente scannerizzata nel dettaglio utilizzando il profilometro laser Gocator 2629. I difetti trovati vengono raggruppati e tradotti in traiettorie dell’utensile per l’operazione di finitura superficiale robotica, verificata attraverso una simulazione in RoboDK ed eseguita con un controllo di feedback in forza. La calibrazione tra i due robot collaborativi Techman TM12 ha garantito la coerenza spaziale tra i sistemi di riferimento, con un errore medio complessivo di circa 3 mm, pienamente compatibile con le tolleranze operative richieste per le lavorazioni automatiche. I risultati sperimentali hanno confermato la capacità del sistema di riconoscere difetti reali e artificiali con elevata accuratezza spaziale e di mantenere costante la forza di contatto durante la finitura, dimostrando una corrispondenza precisa tra traiettorie simulate e reali. Nel complesso, il sistema sviluppato dimostra la fattibilità di una cella robotica completamente automatizzata per l’ispezione e la finitura superficiale, in grado di ridurre l’intervento umano garantendo al contempo precisione, ripetibilità ed efficienza. Il lavoro rappresenta un contributo concreto verso lo sviluppo di sistemi robotici intelligenti e adattivi per l’identificazione e la correzione dei difetti, pienamente in linea con i principi della manifattura 4.0.
Dual-robot system for 3D reconstruction, surface defect detection and defect removal path planning on unknown complex freeform surfaces
Volpi, Riccardo
2024/2025
Abstract
This thesis presents the design, implementation and validation of an integrated robotic system for the automated inspection and surface finishing of complex surfaces, based on the combination of a line laser scanner, an RGB-D camera and two collaborative robots. The goal was to develop a complete and autonomous workflow able to reconstruct the 3D geometry of a component, detect surface defects using robust algorithms and perform automatic correction of imperfections through robotic finishing operations with force feedback control. The proposed system integrates high-resolution optical profilometry with robotic manipulation, enabling accurate acquisition of the surface topography and subsequent correction of defects without human intervention. The initial reconstruction of the component, performed using the RGB-D camera, allows the areas of interest to be selected, which are then analyzed in greater detail by the Gocator 2629 laser profilometer. The defect identification is based on a mobile windows RANSAC interpolation algorithm, capable of robustly isolating local geometric anomalies even in the presence of noise. The detected defects are then grouped and transformed into operational trajectories for robotic sanding, validated in the RoboDK simulation environment and executed via realtime force control. The system calibration between the two Techman TM12 robots ensured coherent transformation of reference frames, achieving an average positional error of approximately 3 mm, fully compatible with the tolerances required for automated finishing. The results confirmed the system’s ability to detect real and artificial defects with high spatial accuracy and to maintain stable contact forces during finishing, demonstrating consistency between simulated and real trajectories. Overall, the developed system demonstrates the feasibility of a complete automatized robotic cell for inspection and surface finishing, able to reduce the human intervention while guaranteeing at the same time precision, repeatability and efficiency. The thesis work represents a concrete contribution towards adaptive and intelligent robotic systems for the identification and correction of defects, aligned with the principles of Industry 4.0.| File | Dimensione | Formato | |
|---|---|---|---|
|
2025_12_Volpi_Tesi.pdf
accessibile in internet per tutti
Descrizione: Tesi Volpi
Dimensione
43.46 MB
Formato
Adobe PDF
|
43.46 MB | Adobe PDF | Visualizza/Apri |
|
2025_12_Volpi_Executive Summary.pdf
accessibile in internet per tutti
Descrizione: Executive Summary Volpi
Dimensione
3.53 MB
Formato
Adobe PDF
|
3.53 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/10589/246426