The OASys Framework

The setting of my PhD research is a long-term research project on autonomous systems (ASys), which addresses a universal technology for the development of all classes of autonomous systems, regardless of its particular application domain. Autonomous systems refer to systems capable of operating in a real-world environment without any form of external control for extended periods of time. The core strategy is to exploit cognitive control loops, using knowledge captured as different models, based on the ontology for autonomous systems (OASys) developed in this research.

An ontology is a formal, explicit specification of a shared conceptualisation, i.e., a machine-readable abstract model where relevant concepts and relationships are identified and explicitly defined by consensus in a group. Ontologies have been widely applied for software engineering. They have also been used as representational mechanisms based on a computational language, to clarify and share the domain knowledge.

OASys captures and exploits the concepts to support the description and the engineering process of any autonomous system, as a domain-ontology structured in two levels of abstraction, innerly organised into Subontologies and Packages. The METHONTOLOGY methodology has been used as design method, with a final implementation in UML.

The autonomous system´s description has been formalised in the ASys Ontology, divided into the System Subontology and the ASys Subontology. The System Subontology contains the elements to define any system, consisting of: the General Systems Package to characterise any system's structure and behaviour based on the General Systems Theory; the Mereology Package with taxonomies of the whole-part concepts and relationships; the Topology Package for topological connections. The ASys Subontology specialises the former concepts for autonomous systems: the Perception Package to describe the perceptive and sensing processes; the Knowledge Package to characterise the different kinds of knowledge used; the Thought Package with task-oriented processes concepts; the Action Package about the operations and performing actors, and the Device Package to detail the devices.

The autonomous system´s engineering process has been formalised in the ASys Engineering Ontology, structured in the System Engineering Subontology and the ASys Engineering Subontology. The System Engineering Subontology gathers the concepts related to an engineering process, based on different metamodels and specifications: the Requirement Package to specify system's requirements; the Perspective Package with viewpoints in a system development; the Engineering Process Package to describe an engineering development process; and the Model-Driven Package to conceptualise the model-driven engineering approach. The ASys Engineering Subontology contains the specialisation and additional ontological elements to describe an autonomous system's generic engineering process: the ASys Requirement Package to characterise process and system quality requirements; the ASys Perspective Package to design an autonomous system from different aspects; the ASys Engineering Process Package to describe a generic autonomous system engineering process.

OASys has been complemented with the development of the OASys-based methodology to exemplify the use of OASys in a generic autonomous system engineering process.

The OASys Framework (ontology + methodology) has been applied in the Robot Control Testbed (RCT), and the Process Control Testbed (PCT). RCT is a collection of mobile robot systems, with a wide range of implementations and capabilities (from conventional SLAM based mobile robots to virtual ones inspired in rat brain neuroscience). PCT involves the development of a robust control architecture for a chemical reaction system (with multiple steady states), providing the system with cognitive capabilities to carry out complex tasks such as fault diagnosis, alarm management, and control system reconfiguration.

Ontology Resources

This section is intended as a minor source of ontology-related material. It is not a comprehensive list (there is much out there!!) but rather a glimpse at what an ontology is and what is going on regarding ontology research...

Take it as a starting point and feel free to look for additional materials ....

Note: Links will only work for ASLab researchers -due to dissemination restrictions- but the files can be found at author's sites.

WHAT IS AN ONTOLOGY

There is not an unique answer to this question. It is a word from Ancient ages (i.e. Greeks and Romans) which means the philosophy of being (ontos = being, logos = treatise). It has evolved from the philosophy domain to be used within the computer science domain.

Just to entertain yourself ....

• The importance of being an O: ontology vs. Ontology (Guarino95) A classical definition you should know to be taken seriously in the ontology world: explicit specification of a conceptualization (Gruber93)
• An extension of Gruber's definition clarifying some aspects (Studer98)
• A rather old comprehensive introduction to ontologies (Uschold96)
• A further definition and the role of ontologies (Uschold99)
• Principles to design an ontology (clarity, coherence, extendability, etc) (Gruber93)

METHODOLOGIES

To develop an ontology is not easy (so I have been told!!) neither from scratch nor using an existing one. Wise people within the ontology domain have proposed some methodologies (aka steps to be followed). If we have not got a better idea, it is worthwhile using one as starting point ...

Some ideas for ontology's freshers:

• A guide to creating your first ontology (Noy01)
• A thorough sum up work on ontologies, methodologies and re-engineering (Fdez-lope02)
• A rather recent (2003, but two years is a long period in research!) and excellent review and comparison on ontology methodologies (Corcho03)
• One of my favourites: METHONTOLOGY (Gomez-perez98)

TOOLS

To help you out with building up an ontology, a myriad of projects and researchers have been developing generic and specific tools for ontology development, merge (take a couple of ontologies and let them share concepts!), annotation and querying.

It would be difficult to show all the available tools. There are some ongoing projects and developments, even so, check on the following:

• A good approach could be reading the following paper (sometimes too Semantic Web biased) (D13)

LANGUAGES

In the end, you will need some kind of ontology language to give structure to your ontology. Several languages have been developed and used throughout the past years. Some were mere research efforts whereas others have evolved (mainly thanks to being used in the Semantic Web).

You can research in more detail a particular language by tracking its references or homepage, however to start with, a good summary is given in (Corcho00)

Getting to the focus of my research, just some hints on what is available regarding agents, autonomous systems and conceptual modelling:

ONTOLOGY FOR AGENTS

There is plenty of available material on research regarding ontologies for agents (mainly within the Semantic Web or applied to information systems). The underlying idea is to have several agents interacting and communicating. To do so, they need to share some common concepts, language or ideas (so to say) usually regarded as an ontology.

Take a look at the following papers: (Colomb02, Colomb02b, Ferrario04).

And if one ontology was not enough, someone proposes a second and third level ontology ... (Sloman05)

ONTOLOGY FOR AUTONOMOUS SYSTEMS

Being the main topic of my research, I find it useful to look for any research on ontologies for autonomous systems. Some papers came up. However, what the rest of the world regards as an autonomous system does not fulfil my own view of the problem. Even so, it is worth it to consider others views...

• Autonomous systems as autonomous information system (Colomb02)
• Real time intelligent control (Evans02)
• More on autonomous systems (Wray04)

ONTOLOGY-BASED SOFTWARE

• Hope ontology-based software does not end as another buzz word. So far, it is trendy and looks great. Is it the track to follow?
• Make your mind up by reading (Stojanovic04, Almeida03,)

There are other several topics not considered in the aforementioned papers. Ontological commitments, Problem Solving Methods (PSM) (ontologies as knowledge modelling and psm as reasoning mechanisms), specific applications, ontologies and UML projects, etc.

A wide and challenging world is out there!!!