|
Pierre maret and Ken sasaki, University of Tokyo
|
tarix | 30.01.2018 | ölçüsü | 444 b. | | #41434 |
|
Autonomous players Autonomous players Intensive communication task No centralization Peripherics - Personal assistants
- Sensors (wearable and fixed)
Each player is an agent Each player is an agent - Autonomy
- Intensive communication
Agents have two components (Agent Oriented Abstraction): - Knowledge: concept classes, instance, actions (that can be further specialized). = Ontology.
- A decision mechanism (associated to utility function): for instance Evaluate a message, Send an Inform, …
Agents are provided with a layer for acting within knowledge communities Agents are provided with a layer for acting within knowledge communities Knowledge community: - a leader + a topic
- dynamic, no concrete existence, extends the topic
- Related actions: create, join, inform, request, leave..
Exchanges are based on contents Example - A: creates a community on concept “Metro station name”
- B: decides to join the community and informs about “Ginza”, instance of “Metro station name”
In our approach User-oriented peripherics are associated to Personal Agents Sensor are associated to Context Agents
Sensor: foot pressure Sensor: foot pressure Context agent produces knowledge: user’s activity level Context agent delivers knowledge: station name Personal assistant of traveler : Personal agent - knows the desired station and a evaluation rule when to wake-up the traveler (activity is low and desired station is reached)
- is interested in “Activity level” and “Metro station names”
Wake up the user if necessary
Application design is made easier Application design is made easier - Agents are made independently
System is open-ended and compliant with pervasive systems
Dostları ilə paylaş: |
|
|