Introduction to

Multi-agent Systems
by
Budditha Hettige
Department of Computer Science
Overview
• What is an Agent?
• Agent properties and characteristics
• What are Multi Agent Systems?
• Conventional vs Multi-agent Software
Paradigms
• Agent-Based Software Applications
• Multi-agent Development Frameworks
• Design your own Multi Agent System

2
What is an Agent?
Agent is used frequently nowadays in:
–Sociology, Biology
–Cognitive Psychology
–Social Psychology
–Computer Science

3
What is an Agent?
• An agent is anything that can be viewed as
perceiving its environment through sensors and
acting upon that environment through effectors
• Human Agent
– Eyes, ears, and other organs for sensors, and hands,
legs, mouth, and other body parts for effectors
• A robotic agent
– Substitutes cameras and infrared range finders for
the sensors and various motors for the effectors
• Software agent
– has encoded bit strings as its percepts and actions
An Agent
Agent definitions
• “An agent is an entity that senses its
environment and acts upon it”
(Russell, 1997)
• “Intelligent agents are software entities that
carry out some set of operations on behalf of a
user or another program, with some degree of
independence or autonomy, and in so doing,
employ some knowledge or representation of
the user’s goals or desires.”
Introduction to Agent Technology (the IBM Agent)6
Agent properties
Autonomy
Agents operate without the direct
intervention of humans or others, and have
some kind of control over their actions and
internal state.

Introduction to Agent Technology 7

Agent properties
Reactivity:
Agents perceive their environment and
respond in a timely fashion to changes that
occur in it.

Introduction to Agent Technology 8

Agent properties
Social ability:
Agents interact with other agents (and
possibly humans) via some kind of agent-
communication language

(Wooldridge and Jennings, 1995)

Introduction to Agent Technology 9

Agents characteristics
• Act on behalf of a user or another program
(Autonomous)
• Sense the environment and acts upon it
(Reactivity)
• Purposeful action
( pro-activity)

10
Agent Environment

Agent

Sensor Action
Input Output

Environment

11
11
 Multi Agent system
Many entities (agents) in a common
environment

Environment

12
 Comparison of two Software
Paradigms
CONVENTIONAL SOFTWARE MULTI-AGENT SOFTWARE

Hierarchies of programs Large networks of small agents

Sequential processing Parallel processing
Centralized decisions Distributed decisions
Instructions Negotiation
Data-driven Knowledge-driven
Rigidity Adaptation
Pre-programmed behaviour Emergent behaviour

13
Two Software Paradigms
CONVENTIONAL MULTI-AGENT
SOFTWARE SOFTWARE

AGENT

Sequential Process Parallel Process

Introduction to Agent Technology 14

Agent Networks
Swarm4
Swarm 2

Swarm3

Swarm1
Swarm 4

Swarm 3

Swarm2
Introduction to Agent Technology 15
CONVENTIONAL Vs MULTI-AGENT

• 10 grades

A B D A C F A B D D

• How many “A” in the set?

Introduction to Agent Technology 16

Conventional System

A B D A C F A B D D

A B D A C F A B D D

231

Sequential Process
Introduction to Agent Technology 17
Multi-agent System

B A F
A
A
D C D
Who has
D grade “A”?
B

Parallel
Process

Introduction to Agent Technology 18

Agent-Based Software
Applications
Real-Time Scheduling of Taxis
• Real-Time Management
System for over 2,000 taxis in
London
• More than 13,000 orders per
day
• In rush hours 1,500 orders per
hour
• Guaranteed pick up of clients
within 15 min’s
• A wide variety of needs and of
types of vehicles
Source: The Power of Agent-Based Software, George Rzevski
20
Real-Time Scheduling of
Car Rentals
• Real-Time Management
System for the European
operation of one of the
largest rent-a-car companies
in the world
• Sequences of operations
• Delivering & washing cars
• Returning drivers
• No-shows, human errors,
delays
Source: The Power of Agent-Based Software, George Rzevski

21
Real-Time Scheduling of
Seagoing Tankers
• Real-Time Management System for a
London based company managing 10% of
the total seagoing tanker capacity in the
world
• Tankers of 300,000 tons
• 500 cargos per year
• Voyage costs £1million per 45 days
– Typical revenue £2.6million for a voyage
Gulf to US
• Managing queuing before entering Panama
canal
• Managing loading/unloading in ports and
canals
• Frequent changes in market fees
depending on demand
Source: The Power of Agent-Based Software, George Rzevski22
Multi Agent System Development
• Types of Agents • Types of person
– Agents – Man
– Managers – Women
– Activators
– Child
• Message Spaces
– Local Queue • Communication
– Global Message Queue – Speak
• Ontology – Read/write
– Rules • Ontology
– Knowledge
– Knowledge
23
Multi-agent Development
Frameworks
• Frameworks have common standards
– (FIPA platforms and communication languages)
• Frameworks save developer’s time
• Standardization of MAS development
• Examples
– JADE
– JaCaMo
– SeSam
– Madkit
– MaSMT
Introduction to Agent Technology 24
Build your own Multi-Agent
System
• Get clear idea about Problem and solution
• Design a multi agent model
• Select suitable multi-agent system
development framework
• Implement agents, communications
• Implement a way to get solution
• Test and tune-up the system

Introduction to Agent Technology 25

About Problem and solution
• Why multi-agent systems are important to
your solution?

Problem
&
Technology

Introduction to Agent Technology 26

Design an agent model
• How multi agent system solve your problem?
• Design
– Agents
• Type, Task,
– Communication
• Messages,
– Ontology
• Knowledge, rules, actions
– Relation
• Group, rule

Introduction to Agent Technology 27

Example

Introduction to Agent Technology 28

Example

Introduction to Agent Technology 29

Select a framework
• Language support
– Java, C#, Android
• Features
• Example
– JADE
– MaDKIT
– MaSMT

Introduction to Agent Technology 30

Implementation
• Develop agent by agent
• Make communication between agents

Introduction to Agent Technology 31

Implementation
• How agent make the solution
– First solution
– Last solution
– Average solution
– Min/max
– Some other method
• Test and tune-up your solution

Introduction to Agent Technology 32