ATRACO Architecturerelated Issues ATRACO Patras Workshop 29 30

  • Slides: 18
Download presentation
ATRACO Architecture-related Issues ATRACO Patras Workshop 29 -30 Sep C. Goumopoulos (CTI)

ATRACO Architecture-related Issues ATRACO Patras Workshop 29 -30 Sep C. Goumopoulos (CTI)

Integration – 1 st Year prototype ¡ Aim: Feel comfortable after(/at) work ¡ Am.

Integration – 1 st Year prototype ¡ Aim: Feel comfortable after(/at) work ¡ Am. I space: I*space @ Uessex ¡ Task model: 1. Set a comfortable temperature 2. Set a comfortable level of lighting 3. Select favorite Media settings ¡ Purpose: to demonstrate adaptation on the ecology behaviour according to the context and policies of the user ¡ Extension: use also Am. In. Office@CTI to demonstrate task realization (adaptation of ecology configuration and operation) in two different environments (testbeds)

Integration – 1 st Year prototype ¡ ¡ Entities: l The device / service

Integration – 1 st Year prototype ¡ ¡ Entities: l The device / service profiles l Sphere Manager l Fuzzy Systems Based Task Agent l Interaction Agent l Planning Agent l One user Limitations to be discussed tomorrow

Integration – 1 st Year prototype ¡ Ontologies: l Sphere ontology will be created

Integration – 1 st Year prototype ¡ Ontologies: l Sphere ontology will be created by manually combining device / service ontologies with FTA ontology, IA ontology, user profile and privacy ontology (if available) l FTA ontology could be created manually based on sphere ontology or the FTA knowledge base (and hence the local ontology) could be learnt by the FTA agent. As the agent is adapting to the user desires and handles the uncerntainities, it will also update its knowledge base and hence update also the ontology. l IA ontology will be created manually l User profile based on the fuzzy systems based behaviour will be created by FTA

ATRACO System 1. Description of aim as a set of goals each modeled as

ATRACO System 1. Description of aim as a set of goals each modeled as a task model 2. Users / Devices / Services / Am. I space, each having its local ontology 3. Software Modules (Sphere Manager, Ontology Manager) 4. Agents (FTA, PA, IA), each having its local ontology 5. Policies (i. e. privacy, interaction, conflict resolution, etc) 6. Sphere Ontology

Components/Managers ¡ Sphere Manager l l l ¡ manages and dissolves activity spheres; initializes

Components/Managers ¡ Sphere Manager l l l ¡ manages and dissolves activity spheres; initializes an ATRACO system and its components for a specific aim; the sphere manager could be viewed as the ATRACO “OS” whereas a sphere as an ATRACO “process” operates as an event service for the other components Ontology Manager l manages the sphere ontology; aligns and merges local ontologies that correspond to the task model that fulfills the goal

Components/Agents ¡ Fuzzy Systems Based Task Agent (FTA) l realizes specific tasks by using

Components/Agents ¡ Fuzzy Systems Based Task Agent (FTA) l realizes specific tasks by using input by various resources and producing actions; l adapts task-based usage of the ecology to the changing user behaviour, environment conditions and context; l updates user profiles; l maintains its own local knowledge base and ontology

Components/Agents ¡ Planning Agent (PA) l resolves an abstract task hierarchy into concrete tasks

Components/Agents ¡ Planning Agent (PA) l resolves an abstract task hierarchy into concrete tasks using the resources of the ambient ecology; l monitors task progress and goal fulfillment; l may trigger the IA to initialize a kind of adaptive dialogue with the user to get more information for further planning; l maintains its own local knowledge base and ontology

Components/Agents ¡ Interaction Agent (IA) l manages user-system interaction using a mixedinitiative dialogue model;

Components/Agents ¡ Interaction Agent (IA) l manages user-system interaction using a mixedinitiative dialogue model; l adapts man-machine interaction to the user context and behaviour; l helps PA to resolve plans when necessary l updates user profiles; l maintains its own local knowledge base and ontology

Activity Sphere

Activity Sphere

Sphere Ontology ¡ encodes the structure and the state of the sphere components. ¡

Sphere Ontology ¡ encodes the structure and the state of the sphere components. ¡ results from merging the local ontologies of those ecology members that are required to achieve the sphere’s goal; ¡ contains the contextual knowledge necessary to realize the concrete tasks that will lead to goal achievement; ¡ the sphere ontology is re-aligned each time a change in the constituent ontologies happens

Sphere Ontology

Sphere Ontology

SM-PA Interaction

SM-PA Interaction

SM-FTA Interaction

SM-FTA Interaction

SM-IA Interaction

SM-IA Interaction

SM-OM Interaction

SM-OM Interaction

Change in YP

Change in YP

Next steps ¡ ¡ What other services are provided/ consumed by FTA, IA, PA,

Next steps ¡ ¡ What other services are provided/ consumed by FTA, IA, PA, OM, SM components Detailed interaction of the ATRACO components, e. g. , PA-IA, FTA-IA Definition of structures sphere. Ref, aim. Ref, user. Ref, YPRef, TMRef, … Start implementing the components SM, OM (CTI), FTA (UEssex), PA (UUlm), IA (LIMSI), YP/Registry(WP 4), Resource Profiles (All), User Profile(All)