Multi-robot systems are rapidly evolving and, thanks to technological advances, are expected to become an integral part of everyday life. Compared to single robots, fleets offer greater efficiency, flexibility, and fault tolerance. However, like any maturing technology, their management, control, and recovery from error conditions pose complex challenges. To fully exploit these technologies, it is essential to develop advanced and robust control strategies at both the individual and collective levels. The goal is to improve key parameters such as productivity, accuracy, safety, and cost efficiency through the coordination of multiple robots. This coordination relies on essential capabilities such as perception, action, communication, learning, and control. This research focuses on new strategies to make multi-robot systems more effective and applicable in real-world contexts. In particular, we examine distributed control methods that do not require a centralized unit, enabling each robot to make autonomous decisions to optimize overall performance and allowing an operator to interact with the entire system rather than with individual agents. One crucial aspect concerns optimizing the coverage of unknown environments for monitoring and exploration, especially in settings that are inaccessible or hazardous to humans. Traditional coverage control assumes complete knowledge of the environment and the robotic system—an assumption that is often unrealistic. To overcome this limitation, we developed a distributed control approach that accounts for communication constraints among vehicles and with the operator. Another central theme is the robots’ ability to adapt to spatially varying phenomena, detected either directly or through other components of the system. Examples include ambient temperature, target identification, or external data integrated into the system. We propose a method to estimate these phenomena and to modify the system’s behavior dynamically. The model is then extended to scenarios in which the phenomena evolve over time, moving beyond the assumption of static fields and enabling dynamic resource management. Beyond ground-based exploration, we extend the approach to the analysis and monitoring of objects and environments with vertical extent. This study draws inspiration from scanning techniques, structural monitoring, and the 3D reconstruction of elements using specialized sensors. Here again, agent behaviors can be adapted based on the collected information, enhancing operational effectiveness. The proposed strategies make multi-robot systems more reliable and better suited to real-world scenarios, improving their adaptability and operability in dynamic, complex environments. Moreover, experimental work on real vehicles has helped refine these solutions, ensuring greater efficiency and practicality.
I sistemi multi-robot stanno evolvendo rapidamente e, grazie ai progressi tecnologici, si prevede che diventeranno parte integrante della vita quotidiana. Rispetto ai robot singoli, le flotte offrono maggiore efficienza, flessibilità e resistenza ai guasti. Tuttavia, come ogni tecnologia in crescita, la loro gestione, il controllo e il recupero da situazioni di errori pongono sfide complesse. Per sfruttare appieno queste tecnologie, è fondamentale sviluppare strategie di controllo avanzate e robuste, sia a livello individuale che collettivo. L'obiettivo è migliorare parametri chiave come produttività, precisione, sicurezza ed efficienza economica attraverso il coordinamento di più robot. Questo coordinamento si basa su funzionalità essenziali come percezione, azione, comunicazione, apprendimento e controllo. Questa ricerca si concentra su nuove strategie per rendere i sistemi multi-robot più efficaci e applicabili in contesti reali. In particolare, esaminiamo metodi di controllo distribuito che non richiedono un'unità centralizzata, permettendo a ogni robot di prendere decisioni autonome per ottimizzare le prestazioni complessive e consentendo a un operatore di interagire con l'intero sistema piuttosto che con singoli agenti. Uno degli aspetti cruciali riguarda l'ottimizzazione della copertura di ambienti sconosciuti per il monitoraggio e l'esplorazione, in particolare in contesti inaccessibili o pericolosi per l'uomo. Il tradizionale "Coverage Control" presuppone una conoscenza completa dell'ambiente e del sistema robotico, un'ipotesi spesso irrealistica. Per superare questa limitazione, abbiamo sviluppato un approccio di controllo distribuito che tiene conto delle restrizioni di comunicazione tra i veicoli e con l'operatore. Un altro tema centrale riguarda la capacità dei robot di adattarsi a fenomeni spaziali variabili, rilevati direttamente o attraverso altre componenti del sistema. Esempi includono la temperatura ambientale, l'identificazione di obiettivi o dati esterni integrati nel sistema. Abbiamo proposto un metodo per stimare questi fenomeni e modificare dinamicamente il comportamento del sistema. Successivamente, il modello è stato esteso a scenari in cui i fenomeni evolvono nel tempo, superando l'ipotesi di staticità e consentendo una gestione dinamica delle risorse. Oltre all'esplorazione terrestre, abbiamo ampliato l'approccio all'analisi e al monitoraggio di oggetti e ambienti con estensione verticale. Questo studio si ispira alle tecniche di scansione, monitoraggio di strutture e ricostruzione 3D di elementi mediante sensori specifici. Anche in questo caso, il comportamento degli agenti può essere adattato sulla base delle informazioni raccolte, migliorando l'efficacia operativa. Le strategie proposte rendono i sistemi multi-robot più affidabili e adatti a scenari reali, migliorando la loro capacità di adattamento e operatività in ambienti dinamici e complessi. Inoltre, il lavoro sperimentale su veicoli reali ha permesso di affinare le soluzioni, garantendone maggiore efficienza e praticità.
Controllo e coordinamento di sistemi multirobot: dalla teoria alle sperimentazioni sul campo
BELAL, MEHDI
2026
Abstract
Multi-robot systems are rapidly evolving and, thanks to technological advances, are expected to become an integral part of everyday life. Compared to single robots, fleets offer greater efficiency, flexibility, and fault tolerance. However, like any maturing technology, their management, control, and recovery from error conditions pose complex challenges. To fully exploit these technologies, it is essential to develop advanced and robust control strategies at both the individual and collective levels. The goal is to improve key parameters such as productivity, accuracy, safety, and cost efficiency through the coordination of multiple robots. This coordination relies on essential capabilities such as perception, action, communication, learning, and control. This research focuses on new strategies to make multi-robot systems more effective and applicable in real-world contexts. In particular, we examine distributed control methods that do not require a centralized unit, enabling each robot to make autonomous decisions to optimize overall performance and allowing an operator to interact with the entire system rather than with individual agents. One crucial aspect concerns optimizing the coverage of unknown environments for monitoring and exploration, especially in settings that are inaccessible or hazardous to humans. Traditional coverage control assumes complete knowledge of the environment and the robotic system—an assumption that is often unrealistic. To overcome this limitation, we developed a distributed control approach that accounts for communication constraints among vehicles and with the operator. Another central theme is the robots’ ability to adapt to spatially varying phenomena, detected either directly or through other components of the system. Examples include ambient temperature, target identification, or external data integrated into the system. We propose a method to estimate these phenomena and to modify the system’s behavior dynamically. The model is then extended to scenarios in which the phenomena evolve over time, moving beyond the assumption of static fields and enabling dynamic resource management. Beyond ground-based exploration, we extend the approach to the analysis and monitoring of objects and environments with vertical extent. This study draws inspiration from scanning techniques, structural monitoring, and the 3D reconstruction of elements using specialized sensors. Here again, agent behaviors can be adapted based on the collected information, enhancing operational effectiveness. The proposed strategies make multi-robot systems more reliable and better suited to real-world scenarios, improving their adaptability and operability in dynamic, complex environments. Moreover, experimental work on real vehicles has helped refine these solutions, ensuring greater efficiency and practicality.| File | Dimensione | Formato | |
|---|---|---|---|
|
Tesi_MB-1.pdf
accesso aperto
Licenza:
Tutti i diritti riservati
Dimensione
29.9 MB
Formato
Adobe PDF
|
29.9 MB | Adobe PDF | Visualizza/Apri |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/362893
URN:NBN:IT:UNIMORE-362893