What exactly is an agent? James Harland 23rd November, 2009.

Slides:



Advertisements
Similar presentations
ARCHITECTURES FOR ARTIFICIAL INTELLIGENCE SYSTEMS
Advertisements

Intelligent Architectures for Electronic Commerce Part 1.5: Symbolic Reasoning Agents.
Formal Semantics for an Abstract Agent Programming Language K.V. Hindriks, Ch. Mayer et al. Lecture Notes In Computer Science, Vol. 1365, 1997
Interaction Modeling for Testing We would generate the test cases based on our understanding of the interactions that may happen. The source is, again,
Formal Modelling of Reactive Agents as an aggregation of Simple Behaviours P.Kefalas Dept. of Computer Science 13 Tsimiski Str Thessaloniki Greece.
Title: Intelligent Agents A uthor: Michael Woolridge Chapter 1 of Multiagent Systems by Weiss Speakers: Tibor Moldovan and Shabbir Syed CSCE976, April.
CSE 471/598, CBS 598 Intelligent Agents TIP We’re intelligent agents, aren’t we? Fall 2004.
Concrete architectures (Section 1.4) Part II: Shabbir Ssyed We will describe four classes of agents: 1.Logic based agents 2.Reactive agents 3.Belief-desire-intention.
Agent UML Stefano Lorenzelli
The Semantic Web Week 18: Part 4 Introduction to Web Services and Intelligent Web Agents Module Website: Practical.
Agent programming in Dribble from beliefs to goals with plans Birna van Riemsdijk Wiebe van der Hoek John-Jules Ch. Meyer.
Application architectures
A Multi-Agent System for Visualization Simulated User Behaviour B. de Vries, J. Dijkstra.
CSE 471/598 Intelligent Agents TIP We’re intelligent agents, aren’t we? Spring 2004.
Design of Multi-Agent Systems Teacher Bart Verheij Student assistants Albert Hankel Elske van der Vaart Web site
CS5371 Theory of Computation Lecture 10: Computability Theory I (Turing Machine)
BDI Agents Martin Beer, School of Computing & Management Sciences,
Applications of agent technology in communications: a review S. S. Manvi &P. Venkataram Presented by Du-Shiau Tsai Computer Communications, Volume 27,
1 Chapter 19 Intelligent Agents. 2 Chapter 19 Contents (1) l Intelligence l Autonomy l Ability to Learn l Other Agent Properties l Reactive Agents l Utility-Based.
Multi-Agent System for Online Auctions Jose Manuel Valladares Pernas CIS Master Project - Fall 2004 Advisor: Dr. Haiping Xu.
Towards A Multi-Agent System for Network Decision Analysis Jan Dijkstra.
Application architectures
UML Sequence Diagrams Michael L. Collard, Ph.D. Department of Computer Science Kent State University.
Intelligent Agents. Software agents O Monday: O Overview video (Introduction to software agents) O Agents and environments O Rationality O Wednesday:
Principles of Programming Chapter 1: Introduction  In this chapter you will learn about:  Overview of Computer Component  Overview of Programming 
DCT 1123 PROBLEM SOLVING & ALGORITHMS INTRODUCTION TO PROGRAMMING.
Software Engineering 2003 Jyrki Nummenmaa 1 USE CASES In this lecture: Use cases - What are use cases? - Why to use use cases? - How to write.
ARTIFICIAL INTELLIGENCE [INTELLIGENT AGENTS PARADIGM]
1.1 1 Introduction Foundations of Computer Science  Cengage Learning.
Software Engineering – University of Tampere, CS DepartmentJyrki Nummenmaa USE CASES In this lecture: Use cases - What are use.
Intelligent Agents & Agent-oriented systems James Harland School of Computer Science and IT Intelligent Agents & Agent-oriented systems.
An Introduction to Software Architecture
1 Object-Oriented Modeling Using UML (2) CS 3331 Fall 2009.
Building a Robocup team using Jason BDI framework Sean Bassett, Nancy Ho Woo, Laurence Smith, Carlos Valencia Software Agents November 30, 2006.
Introduction to AgentSpeak and Jason for Programming Multi-agent Systems (1) Dr Fuhua (Oscar) Lin SCIS Athabasca University June 19, 2009.
ARTIFICIAL INTELLIGENCE [INTELLIGENT AGENTS PARADIGM] Professor Janis Grundspenkis Riga Technical University Faculty of Computer Science and Information.
Requirements as Usecases Capturing the REQUIREMENT ANALYSIS DESIGN IMPLEMENTATION TEST.
EEL 5937 Agent models. EEL 5937 Multi Agent Systems Lecture 4, Jan 16, 2003 Lotzi Bölöni.
University of Windsor School of Computer Science Topics in Artificial Intelligence Fall 2008 Sept 11, 2008.
Week III  Recap from Last Week Review Classes Review Domain Model for EU-Bid & EU-Lease Aggregation Example (Reservation) Attribute Properties.
October 27, 2006 BDI Agents and AgentSpeak(L) Romelia Plesa PhD Candidate SITE, University of Ottawa.
Fariba Sadri - ICCL 08 Agent Examples 1 Agents Some Examples Fariba Sadri Imperial College London ICCL Summer School Dresden August 2008.
A Quantitative Trust Model for Negotiating Agents A Quantitative Trust Model for Negotiating Agents Jamal Bentahar, John Jules Ch. Meyer Concordia University.
Geoinformatics 2006 University of Texas at El Paso Evaluating BDI Agents to Integrate Resources Over Cyberinfrastructure Leonardo Salayandía The University.
Computer Science 101 Theory of Computing. Computer Science is... The study of algorithms, with respect to –their formal properties –their linguistic realizations.
EEL 5937 Agent communication EEL 5937 Multi Agent Systems Lotzi Bölöni.
Agent Overview. Topics Agent and its characteristics Architectures Agent Management.
Rational Agency CSMC Introduction to Artificial Intelligence January 8, 2004.
Chapter 5 System Modeling. What is System modeling? System modeling is the process of developing abstract models of a system, with each model presenting.
Chapter 10 Algorithmic Thinking. Learning Objectives Explain similarities and differences among algorithms, programs, and heuristic solutions List the.
Do software agents know what they talk about? Agents and Ontology dr. Patrick De Causmaecker, Nottingham, March 7-11, 2005.
1 Specification A broad term that means definition Used at different stages of software development for different purposes Generally, a statement of agreement.
Design and implementation Chapter 7 – Lecture 1. Design and implementation Software design and implementation is the stage in the software engineering.
Automata & Formal Languages, Feodor F. Dragan, Kent State University 1 CHAPTER 3 The Church-Turing Thesis Contents Turing Machines definitions, examples,
MTA SZTAKI Department of Distributed Systems Hogyan mixeljünk össze webszolgáltatásokat, ontológiákat és ágenseket? Micsik András.
Intelligent Agents Chapter 2. How do you design an intelligent agent? Definition: An intelligent agent perceives its environment via sensors and acts.
CS 501: Software Engineering Fall 1999 Lecture 15 Object-Oriented Design I.
Intelligent Agents: Technology and Applications Unit Five: Collaboration and Task Allocation IST 597B Spring 2003 John Yen.
EEL 5937 Multi Agent Systems -an introduction-. EEL 5937 Content What is an agent? Communication Ontologies Mobility Mutability Applications.
Defects of UML Yang Yichuan. For the Presentation Something you know Instead of lots of new stuff. Cases Instead of Concepts. Methodology instead of the.
Multi-Agent System for Online Auctions Jose Manuel Valladares Pernas CIS Master Project - Fall 2004 Advisor: Dr. Haiping Xu.
Service-Oriented Computing: Semantics, Processes, Agents
Designing Agents’ Behaviors and Interactions within ADELFE
Intelligent Agents Chapter 2.
Service-Oriented Computing: Semantics, Processes, Agents
Michael Wooldridge presented by Kim Sang Soon
An Introduction to Software Architecture
Subsuption Architecture
Service-Oriented Computing: Semantics, Processes, Agents
Presentation transcript:

What exactly is an agent? James Harland 23rd November, 2009

What exactly is an agent?23rd November, 2009 Overview  Role of talks  Motivation  Definitions  Proposal  Discussion....

What exactly is an agent? Role of Talks Polished talks on completed work have a role “I am giving this talk at AAMAS/IJCAI/KR/.. next month, so … “ So do informal talks on less developed work x-baked ideas for 0 < x =< 0.5 Informal discussion of developing ideas Opportunities for interactive work This talk is very much an invitation to a discussion … 23rd November, 2009

What exactly is an agent? Motivations Papers at AAMAS’07, ‘08 by ‘Four Musketeers’ on aborting, suspending and resuming goals  Add abort, suspend, resume behaviour to goals and plans  Transform (static) plan library to incorporate abort, suspend, resume -> abort -> [informal] -> -> [formal] 23rd November, 2009

What exactly is an agent? Motivations What exactly is transformed here? Plan library Is that all there is to the agent? In this case, but is that always the case? What exactly needs to be specified to specify an agent? 23rd November, 2009

What exactly is an agent? Agents are... “Usual” features, ie pro-active, reactive, situated, autonomous, social, …. Software (ie an agent is a special case of a program) Analysed in terms of mental attitudes (how do you say “BDI” without saying “BDI”? :-) Need to show both pro-active and reactive behaviours.... (??) 23rd November, 2009

What exactly is an agent? Seeking Agent Definitions... Note that we are looking for a formal definition  Not design principles  Not list of properties  Not (just) examples  Not informal definition (ie English description)  Not #include lots of stuff... None of these are bad, just not what we are looking for … 23rd November, 2009

What exactly is an agent? Agent Definitions Many of the classic papers do not define an agent!  Rao & Georgeff (1991, 1992, …)  Cohen & Levesque (1990)  Shoam (1993)  Franklin & Graesser (1996) .... Exceptions:  Rao's AgentSpeak paper (1996)  Woolridge's 2002 book 23rd November, 2009

What exactly is an agent? AgentSpeak Paper (1996) AgentSpeak(L): BDI Agents Speak Out in a Logical Computable Language, Anand Rao, January, An agent is tuple where E is a set of events (!!) B is a set of beliefs (!!) P is a set of plans (static, ie a plan library) I is a set of intentions (!!) A is a set of actions S E is a function which selects an event from E S O is a function which selects `an option or applicable plan' from a set of applicable plans S I is a function which selects an intention from I 23rd November, 2009

What exactly is an agent? AgentSpeak Paper (1996) Seems to confuse configuration with definition “E,B,I belong in a configuration” “A, P (static), S E, S O, S I in a definition” Rao later defines a BDI configuration as and discusses operational issues 23rd November, 2009

What exactly is an agent? Woolridge's book “We model agents as functions which map runs … to actions” e 0 -> e 1 -> … -> e n Seems accurate, but not too helpful … Later, when describing the architecture, uses: brf: updates beliefs (from beliefs and percepts) options: updates desires (from beliefs and intentions) filter: updates intentions (from beliefs, desires, intentions) plan: selects a plan (from beliefs, intentions, actions) 23rd November, 2009

What exactly is an agent? Multi-agent programming book Chapter 1: JASON and the Golden Fleece of Agent-Oriented Programming, Bordini, Hubner & Veira Agent configuration is ag is an agent program (beliefs and plans) C is a circumstance (intentions, events, actions) M is messages (in and out) and suspended intentions T is temporary information (relevant & applicable plans, chosen intention) s is current step Lots of operational issues. Similar to Rao’s definition. 23rd November, 2009

What exactly is an agent? Multi-agent programming book Chapter 2: Programming Multi-Agent Systems in 3APL, Dastani, van Riemsdijk & Meyer Agent is Configuration is i identifier, σ beliefs, γ goals, CAP capabilities, Π plans, PG goal planning rules PR plan revision rules ξ the environment (!!) θ a substitution 23rd November, 2009

What exactly is an agent? Multi-agent programming book Chapter 3: IMPACT: A Multi-Agent Framework with Declarative Semantics, Dix & Zhang Agent state: “data objects stored in relations the agent handles” Agent program: Set of rules of the form A <- X, L 1, … L n where X is a code call condition, L i are literals 23rd November, 2009

What exactly is an agent? Multi-agent programming book Chapter 4: CLAIM & SYMPA: A Programming Environment for Intelligent and Mobile Agents, El Fallah Seghrouchni & Suna Agent is α name, π parent, K knowledge base G untreated goals, G’ current goals M messages, C capabilities P current running processes S sub-agents, E achieved effects or goals 23rd November, 2009

What exactly is an agent? Agent Example Robots Alpha, Bravo, Charlie are sent to search for survivors of a plane crash. Each is allocated an area to search (perform goal) in the minimum possible time (soft goal) When a survivor is found, the robot gets them back to the base for treatment (achievement goal)  give directions and guidance  ask another robot to help carry the survivor  call for a helicopter Each needs to return to the base every four hours to recharge (maintenance goal) 23rd November, 2009

What exactly is an agent? Agent Example Possible to use a “standard” plan library approach..  Trigger perform goal on start up  Trigger achievement goals when survivors found  Trigger maintenance goals when condition violated (or predicted) What if online planning is used? Must an agent have a plan library? Must part of it be static? 23rd November, 2009

What exactly is an agent? Goal States Pending Maintaining Suspended Active Maintenance Achieve Perform 23rd November, 2009

What exactly is an agent? Agent Example Robots commence with the perform goal and maintenance goal. When a survivor is found, an event is raised, which fires a rule in the plan library which adds the achievement goal When the achievement goal has succeeded, it is dropped. survivor_found: available → +goal(assist_survivor) Maintenance goal is not dropped on success or violation; tends to oscillate between Maintaining and Active states. 23rd November, 2009

What exactly is an agent? Agent Example Why drop achievement goals? Why not send them back to the Pending state? Why use a plan library to trigger goals? Rules for goal transitions (van Riemsdjik, Dastani & Winikoff AAMAS'08) could be used just as easily … 23rd November, 2009

What exactly is an agent? Turing Machines ATuring machine M is Q is a finite set of states Σ is the input alphabet (a subset of Γ\{B}) Γ is the tape alphabet (must contain B) δ partial function from Q x Γ to Q x Γ x {L,R} q 0 is the initial state A Turing machine configuration is S current state T current tape Pos position of tape head pointer Computation is a sequence of configurations defined by the machine and the initial configuration 23rd November, 2009

What exactly is an agent? Turing Machines “Gold standard” for definitions (albeit of a much simpler computational process)  Precise definition of machine  Clear relationship between machine, configuration and computation  Separation of static and dynamic components  Machine definition contains all information needed to determine (the?) next configuration from the current one 23rd November, 2009

What exactly is an agent? How to define an agent? 1. Decide what information is required in an agent configuration 2. Classify each element as static or dynamic Static elements form part of the agent Dynamic elements require  An initial value  A method of update between configurations 23rd November, 2009

What exactly is an agent? So An Agent Should Specify... Agent configuration is 1. Method for updating beliefs  events, percepts, messages,...  learning  self-analysis  … 2. Method for updating goals  adopt/pend/drop/suspend/abort/resume/maintain …  triggered by rules  accepted from other agents  added by user/s  rd November, 2009

What exactly is an agent? So An Agent Should Specify Method for finding plans Plan library On-line planner Other agents … 4. Initial values for all of these 5. Method for plan selection and execution?  Single thread  Concurrent threads  … 23rd November, 2009

What exactly is an agent? Discussion time … 23rd November, 2009