Many robotics applications rely on graph models in one form or another: perception via probabilistic graphical models such as Bayesian Networks or Factor Graphs; control diagrams and other computational ``function block'' models; software component architectures; Finite State Machines; kinematics and dynamics of actuated mechanical structures; world models and maps; knowledge relationships as ``RDF triples''; etc.In traditional graphs, each edge connects just two nodes, and graphs are ``flat'', that is, a node does not contain other nodes. This paper advocates the research hypothesis that hierarchical hypergraphs are a better structural meta model: (i) an edge can connect more than two nodes, (ii) the attachment between nodes and edges is made explicit in the form of ``ports'' to provide a uniquely identifiable view on a node's internal behaviour, and (iii) every node can in itself be another hierarchical hypergraph.These properties are encoded formally in a Domain Specific Language (or ``meta meta model''), called ``NPC4'', built with node, port, connector, and container as primitives, and contains and connects as relationships. The formal model of NPC4 is designed to maximally support its composability as a meta modelling language, for both the structural and behavioural parts of more concrete DSLs that can be built on top of it, each in a specific domain context. NPC4 introduces a particular primitive, the container, to support overlapping contexts. It targets the following major targets in knowledge-centric robotics systems: (i) various levels of abstraction in domain models, (ii) ``multiple inheritance'' from (or rather ``conformance to'') different knowledge domains, and (iii) connecting one or more domain DSLs to the same software infrastructure in which they all have to be ``activated''.

(2016). Hierarchical Hypergraphs for Knowledge-centric Robot Systems: a Composable Structural Meta Model and its Domain Specific Language NPC4 [journal article - articolo]. In JOURNAL OF SOFTWARE ENGINEERING IN ROBOTICS. Retrieved from http://hdl.handle.net/10446/87779

Hierarchical Hypergraphs for Knowledge-centric Robot Systems: a Composable Structural Meta Model and its Domain Specific Language NPC4

2016

Abstract

Many robotics applications rely on graph models in one form or another: perception via probabilistic graphical models such as Bayesian Networks or Factor Graphs; control diagrams and other computational ``function block'' models; software component architectures; Finite State Machines; kinematics and dynamics of actuated mechanical structures; world models and maps; knowledge relationships as ``RDF triples''; etc.In traditional graphs, each edge connects just two nodes, and graphs are ``flat'', that is, a node does not contain other nodes. This paper advocates the research hypothesis that hierarchical hypergraphs are a better structural meta model: (i) an edge can connect more than two nodes, (ii) the attachment between nodes and edges is made explicit in the form of ``ports'' to provide a uniquely identifiable view on a node's internal behaviour, and (iii) every node can in itself be another hierarchical hypergraph.These properties are encoded formally in a Domain Specific Language (or ``meta meta model''), called ``NPC4'', built with node, port, connector, and container as primitives, and contains and connects as relationships. The formal model of NPC4 is designed to maximally support its composability as a meta modelling language, for both the structural and behavioural parts of more concrete DSLs that can be built on top of it, each in a specific domain context. NPC4 introduces a particular primitive, the container, to support overlapping contexts. It targets the following major targets in knowledge-centric robotics systems: (i) various levels of abstraction in domain models, (ii) ``multiple inheritance'' from (or rather ``conformance to'') different knowledge domains, and (iii) connecting one or more domain DSLs to the same software infrastructure in which they all have to be ``activated''.
articolo
Enea, Scioni; Nico, Huebel; Sebastian, Blumenthal; Azamat, Shakhimardanov; Markus, Klotzbuecher; Hugo, Garcia; Herman, Bruyninckx
File allegato/i alla scheda:
File Dimensione del file Formato  
99-581-1-PB.pdf

accesso aperto

Versione: publisher's version - versione editoriale
Licenza: Creative commons
Dimensione del file 1.13 MB
Formato Adobe PDF
1.13 MB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

Caricamento pubblicazioni consigliate

Aisberg ©2008 Servizi bibliotecari, Università degli studi di Bergamo | Terms of use/Condizioni di utilizzo

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/10446/87779
Citazioni
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact