SWARM INTELLIGENCE Sumesh Kannan Roll No 18 Introduction

  • Slides: 17
Download presentation
SWARM INTELLIGENCE Sumesh Kannan Roll No 18

SWARM INTELLIGENCE Sumesh Kannan Roll No 18

Introduction o Swarm intelligence (SI) is an artificial intelligence technique based around the study

Introduction o Swarm intelligence (SI) is an artificial intelligence technique based around the study of collective behavior in decentralized, self-organized systems. o Introduced by Beni & Wang in 1989. o Typically made up of a population of simple agents. o Examples in nature : ant colonies, bird flocking, animal herding etc.

Intelligent Agents o An agent is anything that can be viewed as perceiving its

Intelligent Agents o An agent is anything that can be viewed as perceiving its environment through sensors and acting upon that environment through effectors.

Rational Agents o Rationality - expected success given what has been perceived. o Rationality

Rational Agents o Rationality - expected success given what has been perceived. o Rationality is not omniscience. o Ideal rational agent should do whatever action is expected to maximize its performance measure, on the basis of the evidence provided by the percept sequence and whatever builtin knowledge the agent has. o Factors on which Rationality depends n n Performance measure (degree of success). Percept sequence (everything agent has perceived so far). Agents knowledge about the environment. Actions that agent can perform.

Structure of IA o Agent = Program + Architecture o A Simple Agent Program.

Structure of IA o Agent = Program + Architecture o A Simple Agent Program.

Simple Reflex Agents o Follows Condition-Action Rule. o Needs to perceive its environment completely.

Simple Reflex Agents o Follows Condition-Action Rule. o Needs to perceive its environment completely.

Model Based Agents o Need not perceive the environment completely. o Maintains an internal

Model Based Agents o Need not perceive the environment completely. o Maintains an internal state. o Internal states should be updated.

Goal Based Agents o Makes decisions to achieve a goal. o More flexible.

Goal Based Agents o Makes decisions to achieve a goal. o More flexible.

Utility Based Agents o A complete specification of the utility function allows rational decisions

Utility Based Agents o A complete specification of the utility function allows rational decisions in two kinds of cases. n n Many goals, none can be achieved with certainty. Conflicting goals.

Environment o Accessible vs. Inaccessible o Deterministic vs. Non-deterministic o Episodic vs. Non-episodic o

Environment o Accessible vs. Inaccessible o Deterministic vs. Non-deterministic o Episodic vs. Non-episodic o Static vs. Dynamic o Continuous vs. Discreet

An Environment Procedure

An Environment Procedure

Ant Colony Optimization (ACO) o First ACO system- Marco Dorgo, 1992 o Ants search

Ant Colony Optimization (ACO) o First ACO system- Marco Dorgo, 1992 o Ants search for food. o The shorter the path the greater the pheromone left by an ant. o The probability of taking a route is directly proportional to the level of pheromone on that route. o As more and more ants take the shorter path, the pheromone level increases. o Efficiently solves problems like vehicle routing, network maintenance, the traveling salesperson.

Particle Swarm Optimization (PSO) o Population based Stochastic optimization technique. o Developed by Dr.

Particle Swarm Optimization (PSO) o Population based Stochastic optimization technique. o Developed by Dr. Eberhart & Dr. Kennedy in 1995. o The potential solutions, called particles, fly through the problem space by following the current optimum particles. o Applied in many areas: function optimization, artificial neural network training, fuzzy system control etc.

Swarm Robotics o Most important application area of Swarm Intelligence o Swarms provide the

Swarm Robotics o Most important application area of Swarm Intelligence o Swarms provide the possibility of enhanced task performance, high reliability (fault tolerance), low unit complexity and decreased cost over traditional robotic systems o Can accomplish some tasks that would be impossible for a single robot to achieve. o Swarm robots can be applied to many fields, such as flexible manufacturing systems, spacecraft, inspection/maintenance, construction, agriculture, and medicine work

Applications o Massive (Multiple Agent Simulation System in Virtual Environment) Software. n o Developed

Applications o Massive (Multiple Agent Simulation System in Virtual Environment) Software. n o Developed Stephen Regelous for visual effects industry. Snowbots n Developed Sandia National laboratory.

References http: //en. wikipedia. org http: //www. swarmbots. com http: //www. siprojects. com

References http: //en. wikipedia. org http: //www. swarmbots. com http: //www. siprojects. com

Thank you

Thank you