JACK Intelligent Agents and Applications Hitesh Bhambhani CSE 6362, SPRING 2003 Dr. Lawrence B. Holder
Sunday, June 14, 2015 Contents Introduction to JACK BDI Agents JACK Framework JACK Applications Conclusions and Discussion
Sunday, June 14, 2015 Introduction to JACK Agent Oriented Software, Melbourne, Australia. JACK Intelligent Agents Agent oriented development environment built on top of JAVA Provide developers with a robust, stable, light- weight product Mainly for use as components of larger environments
Sunday, June 14, 2015 Introduction to JACK Features Agent architecture independent facilities Support for BDI agents Agent Communication Language (ACL) independent Geared towards use of middleware and message passing infrastructures SimpleTeam technology to build agent teams
Sunday, June 14, 2015 BDI Agents BELIEFS Description of the state of the world (state as viewed by an agent) DESIRES/GOALS Functions that specify the states of the world that are desirable INTENTIONS/PLANS A set of actions generated by the agent to change the state of its current environment
Sunday, June 14, 2015 BDI Agents Conceptual Model Beliefs Plans ACL Blackboard Goals Environment Beliefs
Sunday, June 14, 2015 JACK Framework Extends Java Keywords to support main concepts e.g. agent, plan, event Statements for declaration of attributes and characteristics of components Statements for defining static relations Statements for agent state manipulation
Sunday, June 14, 2015 JACK Framework JACK Compiler – JACK Agent Language into Java JACK Runtime Environment Execution of JACK agents Handles communications Simple agent naming service
Sunday, June 14, 2015 JACK Framework Agent Development Environment software.com/shared/products/images/Design.gif software.com/shared/products/images/Design.gif Agent Debugging Environment View messages between agents View internal execution states
Sunday, June 14, 2015 JACK Framework SimpleTeam Programming Team oriented programming Agent collaboration viewed as abstract viewpoint of a group as a whole Neutral to nature of team structure Centralized specification of coordinated activity Achieved through mechanisms of team, role, team capabilities and team plans.
Sunday, June 14, 2015 JACK Applications- ROOBOTS The Roobots are a team of robots from the University of Melbourne Entered the small-sized robot league, and finished in fourth place at RoboCup-2002 in Fukuoka, Japan. Roobots Design uses JACK Agents based on BDI obots.html obots.html
Sunday, June 14, 2015 JACK Applications- Decision Support Intelligence Preparation for Battlefield Allocation of surveillance and reconnaissance assets Planners needed to consider following to plan: Number and priority of CCIR (Commander ’ s Critical Information Request) Nature of surveillance needed to satisfy CCIR Assets available to plan with Ongoing availability taken in account priorities, maintenance, etc Re-plan for developing tactical situation
Sunday, June 14, 2015 JACK Applications- Decision Support Collection Plan Management System (CPMS) Planning module based on JACK agents Visualization module containing Geographical Information System Database Terrain data CCIRs Surveillance assets (Capabilities, availabilities, etc)
Sunday, June 14, 2015 JACK Applications- Decision Support(CPMS)
Sunday, June 14, 2015 JACK Applications- Decision Support CPMS planning module Inputs List of CCIRs Order of Battle (ORBAT) Command and control (C2) structure Available assets Current surveillance plan (optional) Outputs Set of alternative plans for the commander
Sunday, June 14, 2015 JACK Applications- Decision Support CPMS planning module implementation Multi-agent subsystem Agent hierarchy same as C2 structure Top level agent prioritizes alternative tasks in CCIRs Ask subordinate agents for plans Each subordinate agent applies interdependency constraints to tasks and generates plan
Sunday, June 14, 2015 JACK Applications – Manufacturing Plant JACK-based holonic manufacturing at the Cambridge Robotic Assembly Cell at Cambridge University, England A Holon is the combination of a physical process and an intelligent agent Demo ( software.com/shared/demosNdocs/flashDemos/LongHolo nicDemo.html) software.com/shared/demosNdocs/flashDemos/LongHolo nicDemo.html
Sunday, June 14, 2015 JACK Applications – Defense Simulation Defense simulations war games - fully scripted scenario where entity activities pre-programmed lacking in modeling humans Wish to model human roles and their reasoning in a Complex Command and Control (C3) system Simulation Agent Infrastructure JACK agents to model human behavior SimpleTeam used to model entity and group roles and behaviors
Sunday, June 14, 2015 Conclusions Framework for industrial strength agents Compatible with various agent architectures and legacy systems Tool support for design and debugging However Not ‘ pure ’ AI Current implementations mostly planning apps
Sunday, June 14, 2015 Questions and Discussion