FIPA | 96/06/26 09:31 |
FOUNDATION FOR INTELLIGENT PHYSICAL AGENTS | Yorktown |
Source: M. Busuioc, H. Christensen, D. Connolly, G. Granlund, K. Karlson, P. Kearney and D. Sadek | nystditems |
This document has been produced at the Yorktown meeting. An ad-hoc group has been established to refine the document until the Tokyo meeting.
The standardisation work is divided into two kinds of
activities
The standardisation work is divided into two kinds of
activities
- Definition of reference applications
- Measurement/quantification of agent characteristics
- assessing the quality of an agent.
- Modes of addressing
* Topic based
* Semantic
* broad-, multi-, anycast
- Model of communication
- Protocols
* QoS negotiation
* Synchronisation and establishment of time-base
- Typical cooperative behaviour principles
- Dialogue (e.g. corrective, suggestive, over-informative,
multi-threaded, answers)
- Rules of agent behaviour (e.g. commitments)
- Error detection and recovery
- Input/output primitives
- Selecting and accessing services
- Speech, natural language and gesture interfaces
- Interactions of agent based subsystems.
- Coordination/negotiation with others
- Communication languages and delivery service, e.g. KQML, KIF,
etc
- Reference framework e.g. definition of domain specific
vocabulary, CORBA
- OpenDoc, ...
- Speech input
- Speech output
- Features for speech recognition
- Identification of objects and determination of their
positions
- Image output
- Graphics output
- Coordination/fusion of multi-modalities
- GPS, Sonars, gyros, ...
- I/O primitives for mechanical actuators, such as Active
Head-Eye Systems
- I/O for manipulation, mobility
- Specification of physical parameters
- Information fusion
- Input and output facial and speech parameters
- Multimodal interaction (speech, gesture, gazing etc.)
- Representation of information by content-based indexing
- Knowledge representation reference models - Rational agents and
human-agent communication:
- Interworking of agent systems, e.g. languages, contents
representation
- Creating and sharing ontologies
- Agent security (e.g. certification, security level,
signalling)
- Transaction security and privacy
- Environment (e.g. other agents, user, etc.) security and
privacy
Needs definition of society vs interagent communication
- Society issues
* Cooperation primitives, e.g. agreement
- Scalability of large agent communities
- Communities of heterogeneous agents
- Rules of agent behaviour (e.g. commitments)
- definition generic agent programming templates
- definition of a set of agent programming primitives
- Agent toolkits incorporating an open design approach to
* development,
* maintenance
* distribution
- Standards for user profile description
- Specialisation agents system behaviours to individual user
preferences
- Mobile execution models
- Configuration issues
- Administration and Policing
- Recommended skills list
- Definition of ontologies (e.g. service taxonomy)
- Directory: A standard for type management and classification
trees is of critical importance for the whole architecture. Bying
and Selling Agents must be able to access the available goods in
a standardized form.
- Product Delivery (storage and transfer): An important issue is
the representation and handling of electronic and physical goods.
The agents have to arrange for the delivery of goods and ensure
that a specific item is not offered in more than one place at a
time.
- dynamic extensibility of complex, distributed services