In software-intensive applications such as IoT, assistive robotics, and smart manufacturing, smart cyber agents play a crucial role by monitoring the physical environment through sensors and making impactful decisions acting on the environment. Within these domains, specifying tasks for a system, involving multi-agents interacting with the environment and other agents (including humans, a safety-critical situation), presents a significant challenge. Typically, these specifications are defined using a pre-defined set of patterns combined by the end user, but this is not sufficient to cover all the real word applications. This thesis addresses this challenge by introducing a Domain-Specific Language (DSL) named LIrAs, which stands for Language for Interactive Agents. LIrAs serves as a general-purpose DSL that enables non-expert users to create customized patterns and use them in specifying missions (i.e., combinations of patterns). The high-level capabilities of LIrAs facilitate the unambiguous definition of patterns. Users can leverage a set of basic building blocks and compose them with synchronization stages and action conditioning. The semantics of this DSL are implemented in Xtext and mapped to Stochastic Hybrid Automata (SHA) in Uppaal, making specifications amenable to formal verification. While in this thesis the DSL is exemplified mainly through examples in the service robotics field, its flexibility allows it to be applied in a broad range of applications, thanks to its domain-agnostic nature.
Gli agenti robotici intelligenti, in applicazioni come: lo IoT, la robotica assistiva e la manifattura intelligente, svolgono un ruolo cruciale. Essi monitorano l’ambiente circostante tramite sensori, prendono decisioni e compiono azioni, influenzando l’ambiente in cui si trovano. In questi scenari, specificare i compiti per un sistema multi-agente (inclusi agenti umani), con gli agenti che interagiscono tra loro e con l’ambiente, rappresenta una sfida per l’ingegneria del software. Solitamente, queste specifiche sono definite dall’utente mediante un insieme predefinito di patterns (i.e., sequenze di azioni), che combinati creano missioni più articolate. Questa tesi introduce un linguaggio DSL, chiamato LIrAs, Language for Interactive Agents. LIrAs funge da DSL generico che consente agli utenti non esperti di creare patterns personalizzati e di utilizzarli per specificare missioni. Le capacità di alto livello di LIrAs agevolano la definizione inequivocabile di patterns, dove gli utenti possono sfruttare un insieme di blocchi base, e combinarli tramite sincronizzazioni e condizionamento delle azioni. La semantica di questo DSL è implementata in Xtext e mappata con SHA in Uppaal, rendendo le specifiche idonee alla verifica formale. Sebbene il DSL in questa tesi sia principalmente esemplificato attraverso esempi nel campo della robotica, la sua flessibilità consente di essere utilizzato in una vasta gamma di applicazioni, grazie alla sua natura agnostica.
High-level specification of multi-agent interaction patterns
Tagliaferro, Alberto
2023/2024
Abstract
In software-intensive applications such as IoT, assistive robotics, and smart manufacturing, smart cyber agents play a crucial role by monitoring the physical environment through sensors and making impactful decisions acting on the environment. Within these domains, specifying tasks for a system, involving multi-agents interacting with the environment and other agents (including humans, a safety-critical situation), presents a significant challenge. Typically, these specifications are defined using a pre-defined set of patterns combined by the end user, but this is not sufficient to cover all the real word applications. This thesis addresses this challenge by introducing a Domain-Specific Language (DSL) named LIrAs, which stands for Language for Interactive Agents. LIrAs serves as a general-purpose DSL that enables non-expert users to create customized patterns and use them in specifying missions (i.e., combinations of patterns). The high-level capabilities of LIrAs facilitate the unambiguous definition of patterns. Users can leverage a set of basic building blocks and compose them with synchronization stages and action conditioning. The semantics of this DSL are implemented in Xtext and mapped to Stochastic Hybrid Automata (SHA) in Uppaal, making specifications amenable to formal verification. While in this thesis the DSL is exemplified mainly through examples in the service robotics field, its flexibility allows it to be applied in a broad range of applications, thanks to its domain-agnostic nature.File | Dimensione | Formato | |
---|---|---|---|
2024_04_Tagliaferro_Thesis_01.pdf
accessibile in internet per tutti
Descrizione: testo tesi
Dimensione
3.39 MB
Formato
Adobe PDF
|
3.39 MB | Adobe PDF | Visualizza/Apri |
2024_04_Tagliaferro_Executive_Summary_02.pdf
accessibile in internet per tutti
Descrizione: executive summary
Dimensione
794.33 kB
Formato
Adobe PDF
|
794.33 kB | 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/10589/218501