multiagent systems and distributed artificial intelligence

38
Multiagent systems and Distributed Artificial Intelligence

Upload: kevyn

Post on 22-Jan-2016

26 views

Category:

Documents


1 download

DESCRIPTION

Multiagent systems and Distributed Artificial Intelligence. Agent?( 智能体). Agent: Intelligent Object Intelligent System with Only one agent A problem solving system by A algorithm or A* algorithm. Multiagent system. Intelligent System with two or more agents—Multiagent system - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Multiagent systems and Distributed Artificial Intelligence

Multiagent systems and Distributed Artificial Intelligence

Page 2: Multiagent systems and Distributed Artificial Intelligence

Agent?(智能体 )

Agent: Intelligent ObjectIntelligent System with Only one agentA problem solving system by A algorithm or A* algorithm

Page 3: Multiagent systems and Distributed Artificial Intelligence

Multiagent system

Intelligent System with two or more agents—Multiagent systemGame Playing System by alpha-beta procedure

Page 4: Multiagent systems and Distributed Artificial Intelligence

Why Multi-agent system?

Difference between systems with one agent and multi-agents?See an example.

Page 5: Multiagent systems and Distributed Artificial Intelligence

An example: Boid

Who designs and controls the behavior of Bird flocks, Fish schools?See a computer model for computer simulation of the behavior of bird flocks, fish schools.

Page 6: Multiagent systems and Distributed Artificial Intelligence

3 rules: Separation Separation:

steer to avoid crowding local flockmates

Page 7: Multiagent systems and Distributed Artificial Intelligence

3 rules: AlignmentAlignment: steer towards the average heading of local flockmates

Page 8: Multiagent systems and Distributed Artificial Intelligence

3 rules: CohesionCohesion: steer to move toward the average position of local flockmates

Page 9: Multiagent systems and Distributed Artificial Intelligence

Neighborhood around an agent

Every agent reacts only to flockmates within a certain small neighborhood around itself. The neighborhood is characterized by a distance and an angle,

Page 10: Multiagent systems and Distributed Artificial Intelligence

NeighborhoodThe neighborhood is characterized by a distance (measured from the center of the boid) and an angle, measured from the boid's direction of flight. Flockmates outside this local neighborhood are ignored. The neighborhood could be considered a model of limited perception (as by fish in murky water)

Page 11: Multiagent systems and Distributed Artificial Intelligence

Computer Simulation to Boids

three dimensional computational geometry of the sort normally used in computer animation or computer aided design. See a demo by Java.Sorry. Can not download it.

Page 12: Multiagent systems and Distributed Artificial Intelligence

obstacle avoidance Obstacle avoidance allowed the boids to fly through simulated environments while dodging static objects.

Page 13: Multiagent systems and Distributed Artificial Intelligence

Demo available? See a demoNo Sorry here.

Page 14: Multiagent systems and Distributed Artificial Intelligence

What can we get from the example?

No Central controller. Every agent: its behavior and the relationship to environments Emergence(突现,涌现 )More examples of emergence.

Page 15: Multiagent systems and Distributed Artificial Intelligence

History of Multiagent systemsAbout late 1970sDistributed Artificial Intelligence (DAI) evolved and diversified rapidly.Research and application field. It brings together and draws on results, concepts, and idea from many disciplines: AI, computer science, sociology, economics, organization and management science, and philosophy.

Page 16: Multiagent systems and Distributed Artificial Intelligence

Definition:DAIDAI is the study, construction, and application of multiagent systems, that is, systems in which several interacting, intelligent agents pursue some set of goals or perform some set of tasks.An agent is a computational entity such as a software program or a robot that can be viewed as perceiving and acting upon its environment and that is autonomous in that its behavior at least partially depends on its own experience.

Page 17: Multiagent systems and Distributed Artificial Intelligence

agent An agent can be affected in its activities by other agents.Agents try to combine their efforts to accomplish as a group what the individuals cannot in the case of cooperation.Agents try to get what only some of them can have in the case of competition.

Page 18: Multiagent systems and Distributed Artificial Intelligence

Why multiagent system?-1Modern computing platforms and information environments are distributed, large, open, and heterogeneous.These often exceed the level of conventional, centralized computing because they require processing of huge amounts of data, or of data that arises at geographically distinct locations.

Page 19: Multiagent systems and Distributed Artificial Intelligence

Why multiagent system?-2They have the capacity to play an important role in developing and analyzing models and theories of interactivity in human societies, and solving problems which it is difficult to solve in conventional method.Many interactive processes among humans are still poorly understood, although they are an integreted part of our everyday life.(There are many things we do not know and we want to know related to multiagent systems)

Page 20: Multiagent systems and Distributed Artificial Intelligence

Major characteristics of multiagent systems

Each agent has just incomplete information and is restricted in its capabilities.System control is distributed;Data is decentralized; andComputation is asynchronous.

Page 21: Multiagent systems and Distributed Artificial Intelligence

Some attributes of multiagent systems - 1

attribute range

agents Number From two upward

Uniformity Homogeneous… heterogeneous.

Goals Contradicting … complementary

Abilities ( sensors, effectors, cognition)

Simple … advanced

Page 22: Multiagent systems and Distributed Artificial Intelligence

Some attributes of multiagent systems - 2

attribute range

Interac-tion

Frequency Low … high

Pattern (flow of data and control)

Decentralized … hierarchical

Variability Fixed … changeable

Perpose Competitive… cooperative

Page 23: Multiagent systems and Distributed Artificial Intelligence

Some attributes of multiagent systems - 3

attribute range

Environment

Predictability Forseeable … unforseeable

Accessibility and knowability

Unlimited … limited

Dynamics Fixed … variable

Diversity Poor … rich

Availability of resources

Restricted … ample

Page 24: Multiagent systems and Distributed Artificial Intelligence

Difference between traditional AI and DAI-1

Traditional AI DAI

Concentrates on agents as

“Intelligent stand-alone systems”,

“Intelligent connected systems”,

Concentrates on Intelligence as

a property of systems that act in isolation.

a property of systems that interact.

Page 25: Multiagent systems and Distributed Artificial Intelligence

Difference between traditional AI and DAI-2

Traditional AI DAI

Concentrates on

Cognitive processes within individuals

Social processes in groups of individuals

Considers systems

Internal reasoning and control.

Reasoning and control is distributed

Page 26: Multiagent systems and Distributed Artificial Intelligence

Difference between traditional AI and DAI-3

Traditional AI DAI

uses Psychology and behaviorism for ideas and inspiration.

Sociology and economics

Page 27: Multiagent systems and Distributed Artificial Intelligence

Reasons to study multiagent systems

Technological and application needs:

Offer a promising and innovative way to understand, manage, and use distributed, large-scale, dynamic, open, and heterogeneous computing and information systems.

Natural view of intelligent systems

Page 28: Multiagent systems and Distributed Artificial Intelligence

Another example: Floys

flocking Artificial creatures.with the social tendency to stick

together

Page 29: Multiagent systems and Distributed Artificial Intelligence

Two behavior rules

1. A rule specifying how to relate to one's own kind.

2. A rule specifying how to relate to strangers

Page 30: Multiagent systems and Distributed Artificial Intelligence

How to relate to one's own kind

Identify two members of your flock that are near to you and try to stay close to them, but not too close.

Page 31: Multiagent systems and Distributed Artificial Intelligence

How to relate to strangers

If you are in your territory: When you spot a stranger go after him, if you are close enough - attack

If you are not in your territory:If local Floys chase you - run away.

Page 32: Multiagent systems and Distributed Artificial Intelligence

Rules of Evolution-1

eFloys evolve sexually, where each eFloy is the descendent of two parents.

Mother and father are selected according to the mechanism of 'Survival of the Fittest by Unnatural Selection'.

Page 33: Multiagent systems and Distributed Artificial Intelligence

Rules of Evolution-2

Fitness is defined by two attributes, energy and safety.

If you are an eFloy, you can gain or lose these during your lifetime, and the more you have, the fitter you are

Page 34: Multiagent systems and Distributed Artificial Intelligence

What influences fitness?-1

Food is energy: each time you bite a stranger, your energy is increased. Your best option is to reach the stranger first, and eat him all by yourself.

If you are a stranger, each time you are bitten, your energy decreases.When your energy ends, you die.

Page 35: Multiagent systems and Distributed Artificial Intelligence

What influences fitness?-2

If you move fast, your energy decreases. The faster you move, the more energy you lose.

If you are close to your neighbors, your safety increases.The closer you are to your neighbors, the more safety points you get

Page 36: Multiagent systems and Distributed Artificial Intelligence

A demo.

Wait please.

Page 37: Multiagent systems and Distributed Artificial Intelligence

English Books:

Artificial Intelligence: A new Synthesis, Nils J, Nilsson, 机械工业出版社, 1999, 9北京

Multiagent Systems: A modern approach to Distributed Artificial Intelligence, Edited by Gerhard Weiss, The MIT Press, Cambridge, Massachusetts, London, English.2000

Page 38: Multiagent systems and Distributed Artificial Intelligence

A demo.

Bigeye.au.tsinghua.edu.cn

人工生命 /其它媒体 /boids/