This thesis focuses on the study and design of a Nonlinear Model Predictive Control (NMPC) applied to an Unmanned Aerial Manipulator (UAM). The system under consideration consists of a planar aerial platform modeled as a free-floating base, equipped with a three-link manipulator connected through revolute joints. This configuration allows for the analysis of the main challenges related to the dynamic coupling between the drone and the manipulator, which are typical of free-floating robotic systems. The work has been entirely carried out in a MATLAB simulation environment, using a multibody dynamic simulator that accurately reproduces the interaction between the aerial base and the robotic arm. On this platform, a nonlinear predictive controller was implemented through CasADi and solved using the IPOPT nonlinear optimization solver, with the aim of handling kinematic and dynamic constraints and assessing the system’s behavior in complex operating conditions. A specific focus was placed on evaluating the robustness of the controller in the presence of model mismatch, i.e., the discrepancy between the simulator (plant) model and the prediction model used within the NMPC. This methodological choice made it possible to analyze the controller’s ability to preserve its performance despite differences between the predictive model and the actual system dynamics.
La tesi affronta lo studio e la progettazione di un controllore Nonlinear Model Predictive Control (NMPC) applicato a un manipolatore aereo (UAM, Unmanned Aerial Manipulator). Il sistema considerato è costituito da una piattaforma aerea planare, modellata come base libera, su cui è montato un manipolatore a tre bracci connessi da coppie rotoidali. Tale configurazione consente di analizzare le principali problematiche legate all’accoppiamento dinamico tra drone e manipolatore, tipiche dei sistemi a base flottante. Il lavoro è stato condotto interamente in ambiente MATLAB, utilizzando un simulatore dinamico multibody che riproduce in modo dettagliato le interazioni tra base e braccio robotico. Su tale piattaforma è stato implementato un controllore predittivo non lineare sviluppato tramite CasADi e risolto con il solutore IPOPT, con lo scopo di gestire vincoli cinematici e dinamici e di valutare la risposta del sistema in scenari complessi. Particolare attenzione è stata dedicata all’analisi della robustezza del controllore nei confronti del model mismatch, ovvero della differenza tra il modello del simulatore (plant) e quello utilizzato all’interno dell’NMPC come modello di predizione. Questa scelta metodologica ha permesso di studiare in modo più realistico il comportamento del controllore e la sua capacità di mantenere prestazioni adeguate anche in presenza di discrepanze tra modello e sistema reale.
Sviluppo di un controllore NMPC per manipolazione aerea
BRUNELLO, ANDREA
2024/2025
Abstract
This thesis focuses on the study and design of a Nonlinear Model Predictive Control (NMPC) applied to an Unmanned Aerial Manipulator (UAM). The system under consideration consists of a planar aerial platform modeled as a free-floating base, equipped with a three-link manipulator connected through revolute joints. This configuration allows for the analysis of the main challenges related to the dynamic coupling between the drone and the manipulator, which are typical of free-floating robotic systems. The work has been entirely carried out in a MATLAB simulation environment, using a multibody dynamic simulator that accurately reproduces the interaction between the aerial base and the robotic arm. On this platform, a nonlinear predictive controller was implemented through CasADi and solved using the IPOPT nonlinear optimization solver, with the aim of handling kinematic and dynamic constraints and assessing the system’s behavior in complex operating conditions. A specific focus was placed on evaluating the robustness of the controller in the presence of model mismatch, i.e., the discrepancy between the simulator (plant) model and the prediction model used within the NMPC. This methodological choice made it possible to analyze the controller’s ability to preserve its performance despite differences between the predictive model and the actual system dynamics.| File | Dimensione | Formato | |
|---|---|---|---|
|
Brunello_Andrea.pdf
accesso aperto
Dimensione
1.4 MB
Formato
Adobe PDF
|
1.4 MB | Adobe PDF | Visualizza/Apri |
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/20.500.12608/94272