Autonomous robots are increasingly deployed in emergency scenarios, including disaster response, unsafe environments, and rescue operations. In such contexts, Cooperative Perception (CP) overcomes the limitations of single-robot sensing, such as occlusions and limited range visibility, by allowing multiple agents to share and fuse sensor data in real time. This thesis presents the design and evaluation of a communication framework supporting cooperative perception in a multi-robot system. The proposed architecture leverages the Robot Operating System 2 (ROS2) publish/subscribe model to enable the exchange and fusion of LiDAR data across multiple mobile robots. A dedicated testbed, comprising ROSbot XL platforms equipped with RPLiDAR sensors and Vicon motion capture localization, emulates an emergency-like environment with both static and dynamic obstacles and occlusions. A central node aggregates distributed sensory inputs into a unified map, which is subsequently employed for dynamic path planning and collision avoidance. A series of experiments is conducted in both simulated and real-world environments to evaluate system performance under different conditions. Travel time serves as the primary performance metric, while two key variables are systematically adjusted: the number of robots operating within the environment and the publication frequency of the shared map. The experimental results indicate that increasing the number of robots enhances cooperative efficiency, and that higher map publication frequencies improve the timeliness of information exchange. Together, these factors contribute to a notable reduction in overall travel time, underscoring the benefits of multi-robot coordination and frequent map updates.
I robot autonomi vengono sempre più spesso impiegati in scenari di emergenza, in risposta a disastri, ambienti ostili e operazioni di soccorso. In tali contesti, la Cooperative Perception (CP) supera i limiti della percezione basata su un singolo robot, quali le occlusioni e la ridotta visibilità, consentendo a più agenti di condividere e integrare i dati sensoriali in tempo reale. Questa tesi presenta la progettazione e la valutazione di un framework di comunicazione a supporto della percezione cooperativa in un sistema multi-robot. L'architettura proposta sfrutta il modello publish/subscribe del Robot Operating System 2 (ROS2) per abilitare lo scambio e la fusione di dati provenienti da sensori LiDAR tra diversi robot mobili. È stato sviluppato un ambiente di test dedicato, composto da piattaforme ROSbot XL equipaggiate con sensori RPLiDAR e con il sistema di localizzazione Vicon basato su motion capture, al fine di emulare un ambiente di emergenza con ostacoli statici e con condizioni di occlusione. Un nodo centrale aggrega gli input sensoriali distribuiti in una mappa unificata, successivamente utilizzata per la pianificazione dinamica dei percorsi e per evitare collisioni. Una serie di esperimenti è stata condotta sia in ambienti simulati che reali per valutare le prestazioni del sistema in differenti condizioni operative. Come metrica principale è stato adottato il tempo di percorrenza, mentre due variabili chiave sono state sistematicamente modificate: il numero di robot presenti nell’ambiente e la frequenza di pubblicazione della mappa condivisa. I risultati sperimentali evidenziano che un numero maggiore di robot incrementa l'efficienza cooperativa e che frequenze di pubblicazione più elevate migliorano la tempestività dello scambio informativo. Nel complesso, questi fattori contribuiscono a una significativa riduzione del tempo di percorrenza, sottolineando i vantaggi derivanti dal coordinamento multi-robot e dagli aggiornamenti frequenti della mappa.
Mappatura cooperativa in tempo reale per robot autonomi in ambienti occlusi
PIERGUIDI, MATTEO
2024/2025
Abstract
Autonomous robots are increasingly deployed in emergency scenarios, including disaster response, unsafe environments, and rescue operations. In such contexts, Cooperative Perception (CP) overcomes the limitations of single-robot sensing, such as occlusions and limited range visibility, by allowing multiple agents to share and fuse sensor data in real time. This thesis presents the design and evaluation of a communication framework supporting cooperative perception in a multi-robot system. The proposed architecture leverages the Robot Operating System 2 (ROS2) publish/subscribe model to enable the exchange and fusion of LiDAR data across multiple mobile robots. A dedicated testbed, comprising ROSbot XL platforms equipped with RPLiDAR sensors and Vicon motion capture localization, emulates an emergency-like environment with both static and dynamic obstacles and occlusions. A central node aggregates distributed sensory inputs into a unified map, which is subsequently employed for dynamic path planning and collision avoidance. A series of experiments is conducted in both simulated and real-world environments to evaluate system performance under different conditions. Travel time serves as the primary performance metric, while two key variables are systematically adjusted: the number of robots operating within the environment and the publication frequency of the shared map. The experimental results indicate that increasing the number of robots enhances cooperative efficiency, and that higher map publication frequencies improve the timeliness of information exchange. Together, these factors contribute to a notable reduction in overall travel time, underscoring the benefits of multi-robot coordination and frequent map updates.| File | Dimensione | Formato | |
|---|---|---|---|
|
Pierguidi_Matteo.pdf
accesso aperto
Dimensione
8.99 MB
Formato
Adobe PDF
|
8.99 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/20.500.12608/95452