JACK Intelligent AgentsTM: Summary of an Agent Infrastructure

About This Presentation

Title:

JACK Intelligent AgentsTM: Summary of an Agent Infrastructure

Description:

... of Melbourne and RMIT University to draw upon the resources of two world ... JACOB (Object modelling) user documentation and tutorials. application examples ... – PowerPoint PPT presentation

Number of Views:30

Avg rating:3.0/5.0

Slides: 16

Provided by: michaelw67

Category:

more less

Transcript and Presenter's Notes

Title: JACK Intelligent AgentsTM: Summary of an Agent Infrastructure

1
JACK Intelligent AgentsTMSummary of an Agent
Infrastructure

Dr. Michael Winikoff, Research Scientist
Agent Oriented Software Pty. Ltd
michael.winikoff_at_agent-software.com
http//www.agent-software.com
Also Research Fellow at RMIT University,
Australia
winikoff_at_cs.rmit.edu.au, http//www.cs.rmit.edu.au
/winikoff

Autonomous Agents 2001.ppt
2
Agent Oriented Software Pty. Ltd.

Privately owned company based in Melbourne with
offices also in Adelaide and Cambridge (UK)
Founded in 1997 by a group of professionals
experienced in distributed, OO and agent systems
Business objectives
development and marketing of components based on
intelligent agents
contract research and development of advanced
software systems incorporating novel
technologies, including intelligent software
agents
consulting services for development and
deployment of advanced, distributed, real-time
software solutions

3
Agent Oriented Software Pty. Ltd.

One of the few companies worldwide with
capabilities in the rapidly developing technology
of intelligent agents
Collaborating with the University of Melbourne
and RMIT University to draw upon the resources of
two world leading research departments in
intelligent agent software
Customers include UK Ministry of Defence,
Canadian Defence Department, Alcatel France,
Australian Department of Defence
JACK used in Universities world-wide

4
Agents

Situated (in environment)
Autonomous
Reactive
Pro-active
Social
Usually interested in intelligent agents situated
in difficult environments (such as the real
world!)
Can have multiple goals
Can have multiple ways of working toward a goal
Need to make rational choices

Mixture
5
An intelligent agent -the intentional or BDI
agent
Human
Belief, Desire, Intentions Agent
Beliefs - database of perceived world knowledge
Beliefs - perceived understanding of the world
Goals or desires
Goals or desires
Execution Engine
Intentions - currently executing plans
Behaviours - pre-compiled plans
Accumulated experience and behaviours
6
JACK Intelligent Agents?

Based on companys RD
third generation agent system
Entirely written in Java
allows access to Java features and add-ons
Run anywhere PDAs to high end servers
First released October 1998.
Version 3.1 released March 2001
Commercial product documented and support

7
JACK Design Goals

Provide developers with robust, stable, light
weight product
Satisfy a range of practical application needs
Ease technology transfer
Enable further applied research
Design Philosophy
ease of integration, re-use and re-engineering
modular design using plugins, extensible
low resource requirements
software engineering (modules, type checking)

8
JACK Component Architecture
SimpleTeam Model
Application code
BDI agent Model
agent
capability
event
plan
database
view
Utility classes
JACK Kernel
Communication
9
Multi-agency in JACK

Built-in support for messaging (send, reply)
light-weight and fast
simple agent naming scheme
sending and receiving Java objects
decoupled from transport layer, with standard
transport formats including XML, binary and ASCII
encoding
can also use FIPA ACL, KQML, CORBA, RMI, HLA
Agnostic

10
Low Resource Requirements and Scalability

Hundreds of agents running on low-end hardware
Agent creation gt1,000 per second
Agent destruction gt100,000 per second
Message to another agent (in same process)
100,000 per second
JACK agents can be
all concentrated in a single operating system
process
scattered over a multi-platform network
any combination of the previous
High performance component of larger system in
keeping with design philosophy

11
Team support JACK SimpleTeam

A complete team modelling framework
teams and roles, organisation structure
team plans with parallel, coordinated activities
dynamic team formation - task teams, skills based
selection
A type safe programming language
Goals
reduce complexity when implementing distributed
control of coordinated behaviour (e.g. sending a
message to a team as a whole)
model social structure (e.g. C2 in military
context)
Software Engineering

12
Teams and Scalability