In recent years, the autonomous driving sector has witnessed a particularly rapid technological evolution. This is due to the significant advantages they offer in terms of road safety, transport efficiency and user comfort. However, to ensure the safe and reliable operation of such systems, an in-depth testing phase is necessary. On-road tests, although essential, are often expensive and present concrete risks related to the complexity and unpredictability of the real environment. In this context, simulation environments represent an essential tool for the development and validation of autonomous driving algorithms. Simulations allow to replicate realistic scenarios in a controlled virtual environment, significantly reducing the costs and risks associated with physical tests, and accelerating the entire development cycle. This thesis is divided into several phases. First, a comparative analysis is conducted between different simulation environments used for autonomous driving, with particular attention to off-road driving contexts. Then, a modular architecture for a simulator is proposed, useful to ensure an easy replacement or upgrade of components without compromising the overall coherence of the system. A specific focus has been placed on the development of a dynamic model of the vehicle, designed to interact in real time with a virtual three-dimensional environment and with a set of simulated sensors. The 3D environment has been entirely rebuilt from scratch, in order to obtain the highest level of customization and control. Finally, the experimental results obtained through a series of tests carried out in predefined operating conditions are presented, with the aim of evaluating the effectiveness and flexibility of the proposed architecture.
Negli ultimi anni, il settore della guida autonoma ha assistito a un'evoluzione tecnologica particolarmente rapida. Questo è dovuto ai significativi vantaggi che offrono in termini di sicurezza stradale, efficienza dei trasporti e comfort per l’utente. Tuttavia, per garantire un funzionamento sicuro e affidabile di tali sistemi, è necessaria una fase di sperimentazione approfondita. I test su strada, seppur fondamentali, risultano spesso onerosi in termini economici e presentano rischi concreti legati alla complessità e imprevedibilità dell’ambiente reale. In questo contesto, gli ambienti di simulazione rappresentano uno strumento essenziale per lo sviluppo e la validazione degli algoritmi di guida autonoma. Le simulazioni consentono di replicare scenari realistici in un ambiente virtuale controllato, riducendo significativamente i costi e i rischi associati ai test fisici, e accelerando l’intero ciclo di sviluppo. La presente tesi si articola in più fasi. Inizialmente viene condotta un’analisi comparativa tra diversi ambienti di simulazione utilizzati per la guida autonoma, con particolare attenzione ai contesti di guida fuoristrada. Successivamente viene proposta un’architettura modulare per un simulatore, utile a garantire una sostituzione o aggiornamento dei componenti agevolmente senza compromettere la coerenza complessiva del sistema. Un focus specifico è stato posto sullo sviluppo di un modello dinamico del veicolo, progettato per interagire in tempo reale con un ambiente tridimensionale virtuale e con un set di sensori simulati. L’ambiente 3D è stato interamente ricostruito ex novo, al fine di ottenere il massimo livello di personalizzazione e controllo. Infine, vengono presentati i risultati sperimentali ottenuti attraverso una serie di prove effettuate in condizioni operative predefinite, con l’obiettivo di valutare l’efficacia e la flessibilità dell’architettura proposta.
Sviluppo di una piattaforma di simulazione per ADAS Off-Road
Forestieri, Francesco
2024/2025
Abstract
In recent years, the autonomous driving sector has witnessed a particularly rapid technological evolution. This is due to the significant advantages they offer in terms of road safety, transport efficiency and user comfort. However, to ensure the safe and reliable operation of such systems, an in-depth testing phase is necessary. On-road tests, although essential, are often expensive and present concrete risks related to the complexity and unpredictability of the real environment. In this context, simulation environments represent an essential tool for the development and validation of autonomous driving algorithms. Simulations allow to replicate realistic scenarios in a controlled virtual environment, significantly reducing the costs and risks associated with physical tests, and accelerating the entire development cycle. This thesis is divided into several phases. First, a comparative analysis is conducted between different simulation environments used for autonomous driving, with particular attention to off-road driving contexts. Then, a modular architecture for a simulator is proposed, useful to ensure an easy replacement or upgrade of components without compromising the overall coherence of the system. A specific focus has been placed on the development of a dynamic model of the vehicle, designed to interact in real time with a virtual three-dimensional environment and with a set of simulated sensors. The 3D environment has been entirely rebuilt from scratch, in order to obtain the highest level of customization and control. Finally, the experimental results obtained through a series of tests carried out in predefined operating conditions are presented, with the aim of evaluating the effectiveness and flexibility of the proposed architecture.File | Dimensione | Formato | |
---|---|---|---|
2025_7_Forestieri_Executive_Summary.pdf
non accessibile
Descrizione: Executive Summary
Dimensione
4.53 MB
Formato
Adobe PDF
|
4.53 MB | Adobe PDF | Visualizza/Apri |
2025_7_Forestieri_Tesi.pdf
non accessibile
Descrizione: Testo della Tesi
Dimensione
19.74 MB
Formato
Adobe PDF
|
19.74 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/10589/239964