Jak optymalizować funkcjonowanie systemów autonomicznych?  
[How to optimise the functioning of autonomous systems?]

Krzysztof Nowak (IEEE Autonomous Robotics Ontology Working Group & LO VII, Wrocław)


It would appear that the answer to the question "How should we optimise the 
functioning of autonomous systems?" is: "We should not. Such systems should 
self-optimise.  They are after all autonomous, so they do not require external 
assistance." The question, however, arises: how should complex systems operate 
and be optimised?

We do optimise network flows. Some of these networks could be complex, 
dynamic, autonomous.  How could we understand such networks? What could be 
optimised, and how?  The following path is suggested: (1) network flows for 
complex networks; (2) interacting networks; (3) dynamic networks; (4) 
metabolic networks capable of handling perturbations; (5) autonomous networks. 
Could such a path be followed when analysing supply networks or 
socio-technical networks, and would it facilitate analysing, conceptualising 
and optimising them?

More specifically, when complex systems are networks capable of interacting 
with their environments and improving the ways they function, how should we 
understand them?  In simple supply networks, is it just the goods that flow, 
or are there also flows of data, information, decisions, intentions, 
influences?  How can we conceptualise such systems?  Do we need an ontology, 
would it provide a consensus-based understanding, would it facilitate research 
collaboration?  And if an ontology for such systems was proposed, what should 
it look like? 

In this presentation we consider an ontology based on Dennett's intentional 
stance, extended with a metabolic level.  Selections of concepts and 
conceptual relations will be presented. Autonomous, self-modifying and 
self-optimising systems are the systems of the future. But understanding, 
designing, building and maintaining them are all challenging.  Many questions 
have been asked, not many answers given - but steps could be taken towards 
building specific systems involving interdependent networks and capable of 
handling perturbations.