This thesis work introduces a human-centered collaborative painting framework that combines Preference Based Optimization (PBO) with Dynamic Movement Primitives (DMPs) to optimize robot assistance in industrial painting applications. The proposed system enables the operator to carry out the painting process while the robot continuously adapts its behavior in real-time, dynamically adjusting the orientation of the workpiece to align with the operator’s hand movements. The PBO framework utilizes the GLISp algorithm to iteratively fine-tune key control parameters, including execution time, robot responsiveness, and rotation amplification, based on human feedback. Additionally, modifications to the DMPs framework have been implemented to improve the robot’s reactive behavior and adaptability to ergonomic requirements. These enhancements include scaling up the rotations of the human hand and ensuring smooth, natural trajectory generation, ultimately improving the overall interaction between the operator and the robotic assistant. To validate the effectiveness of the proposed approach, six participants performed the painting task while providing evaluations after each attempt regarding various aspects of the process. The results indicate that the adaptive strategy significantly reduces operator effort while optimizing the efficiency and quality of the painting operation.
Questa tesi introduce un framework collaborativo per la pittura industriale basato sull’ interazione uomo-robot, che combina l’Ottimizzazione Basata sulle Preferenze (PBO) con i Dynamic Movement Primitives (DMPs) per ottimizzare l’assistenza robotica nei processi di verniciatura. Il sistema proposto permette all’operatore di eseguire il processo di pittura mentre il robot adatta continuamente il proprio comportamento in tempo reale, regolando dinamicamente l’orientamento del pezzo in modo da allinearsi ai movimenti della mano dell’operatore. Il framework PBO utilizza l’algoritmo GLISp per affinare iterativamente i principali parametri di controllo, tra cui il tempo di esecuzione, la reattività del robot e l’amplificazione della rotazione, basandosi sul feedback umano. Inoltre, sono state apportate modifiche al framework dei DMPs per migliorare il comportamento reattivo del robot e la sua adattabilità ai requisiti ergonomici. Questi miglioramenti includono la scalatura delle rotazioni della mano umana e la generazione di traiettorie fluide e naturali, favorendo un’interazione più intuitiva ed efficace tra l’operatore e l’assistente robotico. Per validare l’efficacia dell’approccio proposto, sei partecipanti hanno eseguito l’operazione di pittura fornendo valutazioni dopo ogni tentativo su diversi aspetti del processo. I risultati ottenuti dimostrano che la strategia adattiva riduce significativamente lo sforzo dell’operatore, ottimizzando al contempo l’efficienza e la qualità dell’operazione di verniciatura.
Preference based optimization for an adaptive human-robot painting application
Ristic, Marko
2024/2025
Abstract
This thesis work introduces a human-centered collaborative painting framework that combines Preference Based Optimization (PBO) with Dynamic Movement Primitives (DMPs) to optimize robot assistance in industrial painting applications. The proposed system enables the operator to carry out the painting process while the robot continuously adapts its behavior in real-time, dynamically adjusting the orientation of the workpiece to align with the operator’s hand movements. The PBO framework utilizes the GLISp algorithm to iteratively fine-tune key control parameters, including execution time, robot responsiveness, and rotation amplification, based on human feedback. Additionally, modifications to the DMPs framework have been implemented to improve the robot’s reactive behavior and adaptability to ergonomic requirements. These enhancements include scaling up the rotations of the human hand and ensuring smooth, natural trajectory generation, ultimately improving the overall interaction between the operator and the robotic assistant. To validate the effectiveness of the proposed approach, six participants performed the painting task while providing evaluations after each attempt regarding various aspects of the process. The results indicate that the adaptive strategy significantly reduces operator effort while optimizing the efficiency and quality of the painting operation.File | Dimensione | Formato | |
---|---|---|---|
2025_04_Ristic_Tesi_01.pdf
accessibile in internet solo dagli utenti autorizzati
Dimensione
33.02 MB
Formato
Adobe PDF
|
33.02 MB | Adobe PDF | Visualizza/Apri |
2025_04_Ristic_ExecutiveSummary_02.pdf
accessibile in internet solo dagli utenti autorizzati
Dimensione
24.01 MB
Formato
Adobe PDF
|
24.01 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/10589/236131