The PuRSUE (Planner for RobotS in Uncontrollable Environments) Framework's purpose is to make the programming and reprogramming of robot applications easier. It also aims to deal with the programming of complex real scenarios, like the ones that can include multiple mobile robots or the presence of not controllable agents. Furthermore, another goal of the PuRSUE framework is to enlarge the number of people able to get in touch with robotic applications even without having the specific background. A new high-level language PuRSUE-ML (Modelling Language), with the aim to model complexity, was developed in the previous work. The PuRSUE framework, through several translations, can automatically synthesize the control strategy and generate the code ready to be deployed into one robot. All the formal rules that define the new language and the parsing phase are well explained in the previous work. Moreover, the PuRSUE framework already presents features that make it suitable in some simple scenarios, for example, a situation in which a single mobile robot is acting in a well-known environment. To know where to start the improvement, a complete analysis of the existing code was done. The main processes are presented in Appendix B. An improvement of the framework was studied and tested. This improvement aims to make the framework suitable for distributed applications. Simple scenarios with at least two robots are taken into consideration. In simulations, these robots can be coordinate without knowing the complete control strategy. So, the main idea behind the partition of the control strategy and the algorithms used to make it feasible will be shown. The main results obtained from the tests are also reported to show the actual behavior once the distributed control strategy is applied. In the end, a more complex scenario was taken into account to show the robots' behavior in presence of a non-controllable agent.
Il framework PuRSUE (Planner for RobotS in Uncontrollable Environments) si propone di semplificare la programmazione e riprogrammazione di robot. Cerca inoltre di ampliare il numero di persone che possono affrontare la sintesi di strategie di controllo, anche per scenari complessi. Questi scenari possono comprendere la presenza di più robot, che necessitano di coordinamento, e di agenti non controllabili. Nonostante sia un framework ancora in via di sviluppo presenta notevoli potenzialità che una volta implementate potrebbero portare ad un utilizzo di robot in nuovi scenari. Un nuovo linguaggio ad alto livello PuRSUE-ML (Modelling Language), si propone di a modellare tali complessità, é stato sviluppato nel lavoro di tesi precedente. Il framework, a seguito di varie traduzioni, riesce a generare automaticamente il codice di controllo da implementare sul robot. Nella tesi su cui si basa questo lavoro, sono specificate tutte le regole su cui il linguaggio è modellato e su cui si basano le traduzioni successive. Sono state implementate alcune funzioni che lo rendono già utilizzabile per semplici applicazioni che presentano un singolo robot controllabile che agisce in un ambiente conosciuto. Per poter sviluppare ulteriormente questo framework si è resa necessaria un'analisi completa ed approfondita del codice esistente, riportata nell' Appendice B. Un miglioramento del framework é stato studiato. Questo miglioramento mira a renderlo distribuito e quindi implementabile su un sistema che presenta più agenti controllabili. Permetterà quindi il coordinamento di più robot mobili rendendoli capaci di perseguire un obbiettivo comune nonostante ogni robot abbia una conoscenza parziale della strategia di controllo. L'idea alla base della separazione della strategia del controllo e tutti gli algoritmi volti a realizzare tale separazione sono presentati. Sono riportati inoltre i principali risultati ottenuti dai test fatti. Questi test mostrano il comportamento simulato in scenari che presentano più di un robot controllabile ,con complessità crescente. Infine è stato preso in considerazione uno scenario più complesso che presenta anche un agente non controllabile.
Improving the code generation mechanism for the PuRSUE framework
BONETTI, VALENTINA
2019/2020
Abstract
The PuRSUE (Planner for RobotS in Uncontrollable Environments) Framework's purpose is to make the programming and reprogramming of robot applications easier. It also aims to deal with the programming of complex real scenarios, like the ones that can include multiple mobile robots or the presence of not controllable agents. Furthermore, another goal of the PuRSUE framework is to enlarge the number of people able to get in touch with robotic applications even without having the specific background. A new high-level language PuRSUE-ML (Modelling Language), with the aim to model complexity, was developed in the previous work. The PuRSUE framework, through several translations, can automatically synthesize the control strategy and generate the code ready to be deployed into one robot. All the formal rules that define the new language and the parsing phase are well explained in the previous work. Moreover, the PuRSUE framework already presents features that make it suitable in some simple scenarios, for example, a situation in which a single mobile robot is acting in a well-known environment. To know where to start the improvement, a complete analysis of the existing code was done. The main processes are presented in Appendix B. An improvement of the framework was studied and tested. This improvement aims to make the framework suitable for distributed applications. Simple scenarios with at least two robots are taken into consideration. In simulations, these robots can be coordinate without knowing the complete control strategy. So, the main idea behind the partition of the control strategy and the algorithms used to make it feasible will be shown. The main results obtained from the tests are also reported to show the actual behavior once the distributed control strategy is applied. In the end, a more complex scenario was taken into account to show the robots' behavior in presence of a non-controllable agent.File | Dimensione | Formato | |
---|---|---|---|
10452039_BonettiValentina.pdf
accessibile in internet per tutti
Dimensione
6.71 MB
Formato
Adobe PDF
|
6.71 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/10589/164623