Presentation is loading. Please wait.

Presentation is loading. Please wait.

Modulo 1LinguaggixCoord&Coop1 Multi-layered Multi-agent Situated System MMASS.

Similar presentations


Presentation on theme: "Modulo 1LinguaggixCoord&Coop1 Multi-layered Multi-agent Situated System MMASS."— Presentation transcript:

1 Modulo 1LinguaggixCoord&Coop1 Multi-layered Multi-agent Situated System MMASS

2 Modulo 1LinguaggixCoord&Coop2 Motivations about space MAS models –do not explicitly consider the spatial structure of agent environment –despite of the fact that Recent results suggest that the topology of agent interaction is critical to the nature of the emergent behavior of the MAS A large class of problems is characterized by unavoidable spatial features: several domains deal with physical space (e.g. localization problems) or a logical space (e.g. information flow in an organizational structure)

3 Modulo 1LinguaggixCoord&Coop3 An example: mutual awareness in coordination Coordination among people is performed through mutual perception, possibly mediated by artificial agents Logical space (categories and their relations) provides a topology to compute mutual awareness Mutual awareness depends on the logical distance among people

4 Modulo 1LinguaggixCoord&Coop4 MMASS A multi-layered situated MAS: agent actions and interactions are strongly dependent on their position in the structured environment –Situated in an heterogeneous environment (multi-layered) in its properties and/or in its structure neighborhoods are not uniform across the space –Composed by heterogeneous agents Different capabilities and behaviors for agents of different types Different sensitivity to external stimuli –Heterogeneous interaction mechanisms ‘Reaction’ among adjacently situated agents ‘Field diffusion’ throughout the spatial structure of agent environment

5 Modulo 1LinguaggixCoord&Coop5 MMASS ancestors Rooted on basic principles of Cellular Automata –intrinsically include the notions of state and spatial structure ===> uniformity –Extension of CA-based models GAMMA (=> Chemical abstract machine) –Chemical metaphor –Inclusion of space topology

6 Modulo 1LinguaggixCoord&Coop6 MMASS and L*MASS Agent behavior  perception-deliberation-action mechanism –Perception of local environment (e.g. adjacent sites, fields) –Action selection according to agent state, position and type –Action execution Language for MASS (L*MASS)  Set of primitives to specify agent actions –intra-agent : trigger() and transport() –inter-agent : emit() and react()

7 Modulo 1LinguaggixCoord&Coop7 MMASS model

8 Modulo 1LinguaggixCoord&Coop8 Multilayered Multi Agent Situated System (MMASS) MMASS  a constellation of interacting Multi Agent Situated Systems (MASS) Construct(MASS 1 … MASS n ) where MASS denotes a Multi Agent Situated System (MASS) –Space: spatial structure of a layer of agent environment –F: set of fields propagating throughout the Space –A: set of situated agents

9 Modulo 1LinguaggixCoord&Coop9 Agent Structured Environment Multilayered space  set of interacting spaces Space: set P of sites arranged in a network Each site p  P is defined by where –P p  P: set of sites adjacent to p –a p  A  {  }: agent situated in p –F p  F: set of fields active in p

10 Modulo 1LinguaggixCoord&Coop10 Fields – at-a-distance and asynchronous interaction Fields are –generated by agents –propagated throughout the space –perceived by other agents –W f : set of field values –Diffusion f : P X W f X P  W f X…XW f –Compose f : W f …XW f  W f –Compare f : W f X W f  {True, False}

11 Modulo 1LinguaggixCoord&Coop11 Situated Agents a  A : (s current state, p current position) T  –  T : set of states that agents can assume –Perception T :  T  [N X W f1 ] …[N X W f|F| ] Perception T (s) = (c T (s), t T (s)) c T (s): coefficient applied to field values t T (s): sensibility threshold to fields An agent perceives a field f i when Compare T (c i T (s)…w fi,t i T (s)) is True –Action T : set of allowed actions for agents of type T 3 2 4 1 4 a

12 Modulo 1LinguaggixCoord&Coop12 Language for MMASS (L*MASS) to express actions

13 Modulo 1LinguaggixCoord&Coop13 trigger(…) –state(s): the agent state is s –perceive(f i ): the field f i is active in p (f i  F p ) and agents of type T in state s can perceive it and (Compare(c i T (s)*w fi, t i T (s)=True) The effect is the change of agent state action: trigger(s,f i,s’) condit: state(s), perceive(f i ) effect: state(s’)

14 Modulo 1LinguaggixCoord&Coop14 transport(…) –position(s): the agent is situated in s –empty(q), near(p,q): q is a site adjacent to p (q  P p ) and no agent is situated in it (q= ) –perceive(f i ): the field f i is active in p and the agent can perceive it The effect is the change of agent position action: transport(p,f i,q) condit: position(p), empty(q), near(p,q), perceive(f i ) effect: position(q), empty(p)

15 Modulo 1LinguaggixCoord&Coop15 emit(…) –state(s): the agent state is s –present(f,p): a field f is active where the agent is situated f  F p The effect is the emission of a new field to be diffused throughout the space action: emit(s,f,p) condit: state(s) effect: present(f, p’) for all p’  P

16 Modulo 1LinguaggixCoord&Coop16 reaction(…) –state(s): the agent state is s –agreed(a p1, a p2,…, a pn ): agents situated in sites {p 1,p 2,…,p n }  P p have previously agreed to undertake a synchronous reaction The effect is the synchronous state change of the involved agents action: reaction(s, a p1, a p2, …, a pn,s’) condit: state(s), agreed(a p1, a p2,…, a pn ) effect: state(s’)


Download ppt "Modulo 1LinguaggixCoord&Coop1 Multi-layered Multi-agent Situated System MMASS."

Similar presentations


Ads by Google