Download - The Direct Data

loutclankedAI and Robotics

Nov 13, 2013 (3 years and 8 months ago)

72 views

CONTENTS


Types of intelligence


Methods used for mobile robots


Computational intelligence in robotics system


Sensory network for perceiving environment


Mobile Robotic system based on a fuzzy controller


Collision avoidance by a fuzzy controller


Case Study of Mirosot Robot


BASIC DEFINITION OF INTELLIGENT
ROBOT:


Robots

that

are

able

to

perceive

the

environment,

make

decision
,

represent

sensed

data

and

can

acquire

and

usefully

apply

knowledge

or

skills

are

called

as

intelligent

robot
.


TYPES OF INTELLIGENCE:



(1) Artificial



(2) Biological


(3) Computational





INTRODUCTION TO
INTELLIGENT ROBOTIC SYSTEM

INTRODUCTION TO METHODS
INTELLIGENT ROBOTIC SYSTEM

METHODS USED FOR MOBILE
ROBOTS:


(1) Subsumption Architecture


(2) Behavior Based Artificial Intelligence (BBAI )


(3) Monalysa architecture


(4) Model based learning


(5) Hierarchical intelligent control





COMPUTATIONAL
INTELLIGENCE IN ROBOTICS
SYSTEMS

Emerging Synthesis of NN, FS and EC:




(1)
NN
:
NEURAL NETWORKS are useful for recognizing




patterns, classifying input and adapting to


dynamic environments by learning.





(2)
FS:

FUZZY SYSTEMS can cope easily with human




knowledge and can be used to perform inference,




but FS do not fundamentally incorporate any




learning mechanism.



(
3)
EC:

EVOLUTIONARY COMPUTATION tunes neural network


and FS. Furthermore, EC has been used for optimizing


the structure of neural network and fuzzy system. The


goal is for an intelligent system to quickly adapt to a


dynamically changing environment.



(1) Skill



(2) Neurodynamics



(3) Recursive “consciousness”.



ARCHITECTURE OF STRUCTURED INTELLIGENCE

Structured Intelligence for A
Robotic System

A SENSORY NETWORK FOR
PERCEIVING ENVIRONMENT


The

robot

recognizes

quantitative

information

of

the

environment
.



Next,

the

robot

perceives

its

external

environment

through

the

interpretation

of

selective

attention

into

the

qualitative

information

by

sensor

fusion/integration

and

focus/release



The

action

comprises

of

reactive

motion

(reflex),

skilled

motion,

primitive

motion

planning,

and

final

motion

planning
.



MOBILE ROBOTIC SYSTEM
BASED ON A FUZZY

CONTROLLER

Target Trace and Collision Avoidance for A Mobile Robotic
System:

SENSING DIRECTION Di FOR


DETECTING OBSTACLES

CANDIDATE PATHS AVOIDING
COLLISION WITH OBSTACLES


COLLISION AVOIDANCE BY A
FUZZY CONTROLLER


A TRIANGULAR MEMBERSHIP FUNCTION CONCERNING THE DISTANCE Xk



BETWEEN THE MOBILE ROBOT AND OBSTACLES

Sensory

Network

for

Mobile

Robots


-

used

to

construct

compact

&

useful

fuzzy

rules
.


-

fuzzy

rules

are

usually

not

generalized,

but

are

to

be

changed

in


different

environments
.


CASE STUDY OF A MIROSOT
ROBOT BASED ON FUZZY
CONTROLLER

DESCRIPTION OF ENVIRONMENT AND ITS OBJECTIVE:

CONTROL STRUCTURE BLOCK DIAGRAM

CASE STUDY


FUZZY MOTION CONTROLLER:

CASE STUDY

Fuzzy Membership Function:

Distance (D
e
)

Angle (θ
e

)

ZE: ZERO

NL: NEGATIVE LARGE

VN: VERY NEAR

NM: NEGATIVE MEDIUM

NE: NEAR

NS: NEGATIVE SMALL

FA: FAR

ZE: ZERO

VF: VERY FAR

PS: POSITIVE SMALL

PM: POSITIVE MEDIUM

PL: POSITIVE LARGE

CASE STUDY

Fuzzy Associative Memory (FAM):



TABLE: 1




TABLE
:

2


Table 1 for left motor which is FAM rule (
NL,VN: ME
)

corresponds to the following fuzzy association:


IF θ
e

= NL

( negative large)

AND

D
e

= VN
(very near)




THEN
VL = ME

(medium)

CONCLUSION


This seminar presents an entire
integrated structure of intelligent
robotic system based on fuzzy
controller


I have focused mainly on the
perception capabilities based on the
sensory network.