The development of autonomous vehicles to perform mission in critical and hazardous environments or situations is a field of study that can be considered fundamental for the whole mankind. This task can be considered very complex from different point of view. Robots have to face and operate in complex environments and scenarios that can be loosely structured and mutable, requiring tailored strategies to navigate. In addition, another level of complexity is given from scenario that considers cooperation and/or coordination between robots, where different vehicles need to build a society, exchange and share information to perform a joint mission. In literature, to manage these kind of complexity, Multi-Agent System (MAS) theory is cited as a tool capable to manage this issue, by modelling autonomous software components (called agent) that, together, can perform activities than one unique entity wouldn’t be able to perform, or with worse performances. These agents can be flexible enough to reason and plan in a dynamic and unstructured environment by means of a higher level of abstraction, which can be provided by means of a formalization. Because of this, agents could exploit formal models to express concepts such as what they are, what they can do, where they are and how they can cooperate to perform a given mission. The dissertation analyses different systems and technologies for distributed intelligence through a review of a wide state-of-the-art and introduces the design of an architecture of a multi-robot infrastructure for the exploration of complex, loosely structured environments by means of the MAS theory. The proposed infrastructure, that is the innovative aspect of this dissertation, has two objectives: to express different aspects of Robot and Multi-Robot systems by means of models and to introduce the layout of a middleware that can equip different kind of robots, interpret the proposed models and manage the whole Multi-Robot System.
Lo sviluppo di veicoli autonomi per lo svolgimento di missioni in ambienti o situazioni rischiose è un campo di studi considerabile come fondamentale per l’intera umanità. Queste sono considerate molto complesse da diversi punti di vista: i Robot necessitano di operare in ambienti molto complessi che possono essere lascamente strutturati e molto dinamici, richiedendo apposite strategie per la loro esplorazione. In aggiunta, in questi ambienti potrebbe essere richiesta la collaborazione o coordinazione tra i robot, facendo sorgere la necessità di stabilire vere e proprie società capaci di portare a termine la missione assegnatagli. In letteratura, per risolvere queste problematiche, viene solitamente citata la teoria dei Sistemi Multi-Agente (MAS), capace di modellare diversi componenti software autonomi (chiamati Agenti) che, insieme, possono svolgere compiti che un'unica entità centralizzata non sarebbe in grado di gestire. Gli agenti si dimostrano flessibili quanto basta per ragionare e pianificare in ambienti dinamici e non strutturati, sfruttando una formalizzazione che consente un alto livello di astrazione delle informazioni. Per ottenere ciò, gli agenti devono essere in grado di sfruttare modelli formali per esprimere e comprendere concetti su chi siano, cosa possano fare, dove si trovino e come possano cooperare. Questa dissertazione analizza diversi sistemi e tecnologie di intelligenza distribuita attraverso una revisione di un ampio stato dell’arte e introduce il design di un’infrastruttura, basata sulla teoria dei MAS, che consente l’esplorazione di ambienti complessi e non strutturati. L’infrastruttura proposta, che è l’argomento innovativo di questa dissertazione, si pone due obiettivi: definire modelli che riguardano diversi aspetti di robot e Sistemi Multi-Robot e introdurre la struttura di un middleware capace di equipaggiare diverse tipologie di robot, elaborare i modelli proposti e gestire l’intero Sistema Multi-Robot.
Design of multi-agent robotic infrastructures for the exploration of complex and non-structured environments
PANEBIANCO, LUCA
2018
Abstract
The development of autonomous vehicles to perform mission in critical and hazardous environments or situations is a field of study that can be considered fundamental for the whole mankind. This task can be considered very complex from different point of view. Robots have to face and operate in complex environments and scenarios that can be loosely structured and mutable, requiring tailored strategies to navigate. In addition, another level of complexity is given from scenario that considers cooperation and/or coordination between robots, where different vehicles need to build a society, exchange and share information to perform a joint mission. In literature, to manage these kind of complexity, Multi-Agent System (MAS) theory is cited as a tool capable to manage this issue, by modelling autonomous software components (called agent) that, together, can perform activities than one unique entity wouldn’t be able to perform, or with worse performances. These agents can be flexible enough to reason and plan in a dynamic and unstructured environment by means of a higher level of abstraction, which can be provided by means of a formalization. Because of this, agents could exploit formal models to express concepts such as what they are, what they can do, where they are and how they can cooperate to perform a given mission. The dissertation analyses different systems and technologies for distributed intelligence through a review of a wide state-of-the-art and introduces the design of an architecture of a multi-robot infrastructure for the exploration of complex, loosely structured environments by means of the MAS theory. The proposed infrastructure, that is the innovative aspect of this dissertation, has two objectives: to express different aspects of Robot and Multi-Robot systems by means of models and to introduce the layout of a middleware that can equip different kind of robots, interpret the proposed models and manage the whole Multi-Robot System.File | Dimensione | Formato | |
---|---|---|---|
Tesi_Panebianco.pdf
accesso aperto
Dimensione
2.79 MB
Formato
Adobe PDF
|
2.79 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/20.500.14242/94467
URN:NBN:IT:UNIVPM-94467