prashu_sem_report by yubenjoseph@gmail.com

VIEWS: 9 PAGES: 29

									                                     Seminar Report „04




               ROBOTIC CONTROL

                       USING

                    FUZZY LOGIC




                      Presented by

                     C.P.PRASANTH

                       01105012



                         2004




Department Of EEE          1         MESCE Kuttipuram
                                                                      Seminar Report „04




                                ACKNOWLEDGEMENT


 I express my gratitude to my guide Mr. Gylson Thomas without whose inspiration this

 paper would not have materialized.



 I am very much indebted to the Head of the Dept, Dr. P.M.S. Nambisan for his
 support and encouragement.


 I am also grateful to the faculty and staff members of EEE Dept., in particular to
 Mrs. Haseena P. Y., for their relentless support. Also all my friends for their words of
 encouragement and invaluable support.
 I sincerely thank every one.




 Above all I thank the Almighty. Without His will my effort and hard work would not

 have borne fruit.




Department Of EEE                           2                        MESCE Kuttipuram
                                                                       Seminar Report „04




              ROBOTIC CONTROL USING FUZZY LOGIC



                                       ABSTRACT




       A mobile robot or an autonomous guided vehicle has to cope with uncertain,
incomplete or approximate information. Moreover it has to identify sudden perceptual
situations to manoeuvre in real time. This paper describe the use of a fuzzy logic system
applied to the control of an autonomous agent .In this each moment is defined by a set of
fuzzy rules based upon the robot position, its sensor values, distance and angle relative to
the target. To prevent the robot from getting stuck by some obstacles, a path memory
system is created, forcing the robot to seek for new alternatives when it gets trapped. This
improves the efficiency of the fuzzy control system and the ability to avoid deadlock
situations.




Department Of EEE                            3                         MESCE Kuttipuram
                                                       Seminar Report „04




                               CONTENTS

     INTRODUCTION                                         1


     WHAT IS FUZZY LOGIC?                                 2


     FUZZY CONTROLLER                                     4


     WHAT SKILLS DO RODOTS NEED?                          6


     FOR ROBOTIC CONTROL                                  7


     SYSTEM DESCRIBTION                                   8
                       INPUT & OUTPUT OF CONTROLLER       11
                       FUZZY SETS                         13
                       MEMBERSHIP FUNCTION                14
                       RULE SETS                          15
                       INFERENCE                          17
                       DEFUZZIFICATION                   19


     ADVANTAGES                                           20


     DISADVANTAGES                                        21


     CONCLUSION                                           22


    REFERENCES
    LIST OF FIGURE
    LIST OF TABLES



Department Of EEE                    4                 MESCE Kuttipuram
                                                                        Seminar Report „04




                                INTRODUCTION


Automatic guided vehicle or mobile robots is an intelligent machine that has intelligence
to determine its motion starts according to the environment conditions. For an AGV to
operate it must sense its environment be able to plan its operations and then act based on
this plan.



             The running environment could be varied such as the path orientation, road
flatness, obstacle position, road surface friction etc. There are great many uncertainties of
what condition will emerge during its operation. Thus a new control method other than
the conventional control method is demanded to manage the response of the whole
system.



             In the last years, fuzzy logic has been applied mobile robot and autonomous
vehicle control significantly. The best arguments supporting fuzzy control are the ability
to cope with imprecise information in heuristic rule based knowledge and sensor
measurements.




             Fuzzy logic can help design robust individual behaviors units. Fuzzy logic
controllers incorporate heuristic control knowledge. It is convenient choice when a
precise linear model of the system to be controlled cannot be easily found. Another
advantage of fuzzy logic control is to use fuzzy logic for representing uncertainties, such
as vagueness or imprecision which cannot be solved by probability theory. Also fuzzy
logic offers greater flexibility to user, among which we can choose the one that best, fits
the type of combination to be performed.




Department Of EEE                            5                         MESCE Kuttipuram
                                                                        Seminar Report „04




                         WHAT IS FUZZY LOGIC?



Fuzzy logic is another class of AI, but its history and applications are more recent than
those of the expert systems (ES). According to George Boole, human thinking and
decisions are based on “Yes / No” reasoning or “1 / 0 ” logic. According to Boolean logic
developed and expert system principles were formatted based on Boolean logic. It has
been argued that human thinking does not always follow crisp “Yes / No ” logic, but is
often vague, quantitative, uncertain, imprecise or fuzzy in nature.




For example in terms of “ Yes / No ” logic, a thinking rule may be

“ If it is not raining AND outside temperature is less than 80 F ,THEN take a sight seeing
trip for more than 100 miles”




In actual thinking it might be

“IF weather is good AND outside temperature is mild THEN take a long sight seeing
trip”.




Based on the nature of human thinking, Lotfi Zadeh, a computer scientist at the university
of California, Berkeley, originated “The Fuzzy Logic” or “Fuzzy Set Theory” in 1965. In
the beginning, he was highly criticized by professional community, but gradually this
emerged as an entirely new discipline of AI. The general methodology of reasoning in




Department Of EEE                            6                         MESCE Kuttipuram
                                                                         Seminar Report „04



fuzzy logic and expert system by “IF.. THEN…” statements or rules are the same,
therefore it is often called “Fuzzy Expert System ”


       A fuzzy logic can help to supplement an ES and it is sometime hybrid with the latter
to solve complex problems. Fuzzy logic has been successfully applied in process control,
modeling, estimation, identification diagnostics, military science, stock market prediction
etc.




Department Of EEE                              7                        MESCE Kuttipuram
                                                                               Seminar Report „04




                            FUZZY CONTROLLER

  A block diagram of a fuzzy control system is shown in figure.




                                                                 Input U(t)              Output Y(t)


Ref input                                                                     O
            Fuzzyfication      Inference       Defuzzyfication                 Process
                               Mechanism

R(t)
                                Rule
                                Base




Figure 1 Fuzzy controller Block Diagram




The fuzzy controller is composed of the following four elements.


1. A Rule base
2. An inference mechanism
3. A fuzzyfication interface
4. A defuzzyfication interface




Department Of EEE                          8                                  MESCE Kuttipuram
                                                                       Seminar Report „04




        A Rule base is a set of IF-THEN rules , which contains a fuzzy logic
quantification of the expert‟s liquistic description of how to achieve good control.


        An inference mechanism , which emulates the expert‟s decision making in
interpreting and applying knowledge about how best to control the plant.


        A fuzzyfication interface ,which converts controller inputs into information that
the inference mechanics can easily use to activate and apply rules.


        A defuzzyfication interface ,which converts the conclusions of the inference
mechanism into actual inputs for the process.




Department Of EEE                           9                         MESCE Kuttipuram
                                                  Seminar Report „04




                   WHAT SKILLS DO ROBOTS NEED ?




Identification
  Object detection and recognition
  What / Who is that ?


Movements
  Obstacle avoidance and homing.
  How do I move safely ?


Manipulation
  Interacting with objects and environment
  How do I change that ?


Navigation
  Mapping and localization
  Where am I ?




Department Of EEE                            10   MESCE Kuttipuram
                                                                        Seminar Report „04




                             ROBOTIC CONTROL


 Intelligent robot must be equipped with at least three types of functions


1. Appropriate perceptual function facilitated by visual, auditory and other sensors to
 allow the robots to perceive its environment.


2. Intelligent information processing function that allows the robots to process the in
 coming information with respects to a given task.


3. Mechanical function which appropriate controls to allow the robot to move and act as
  desired.




       Fuzzy logic can be applied to mobile robot and autonomous vehicle control
significantly. The best argument   supporting fuzzy control is the ability to cope with
imprecise information in heuristics rule based knowledge and sensor measurements.




Department Of EEE                            11                        MESCE Kuttipuram
                                                                    Seminar Report „04




                           SYSTEM DESCRIPTION



      The robot has eight sensors, two to each side (S0 to S7). In order to simplify the
processing, each group of two sensors was converted into one value by using the mean
value of its readings (Figure 2). Therefore, the new sensors arrangement provides four
sensors‟s reading: ahead, behind, left and right.

      The robot has two motors: right and left.




          Figure 2 Positions of sensors and motors




Department Of EEE                            12                     MESCE Kuttipuram
                                                                          Seminar Report „04




 The Khepera simulation environment provides, at any Khepera‟s location, the angle 
between the robot and the axis x. However, the most important angle for this specific
problem is the one between the robot‟s location and its target (). The  angle can be
calculated using the  angle and the robot coordinates (Figure 3).




                                                                   -




      Figure 3    Calculation of the angle between robot and the target position



d = ((xr - x’)2 + (yr - y’)2)1/2

Where d is the distance to the target, (xr , yr) are the robot coordinates, and (x’, y’) are the
coordinates of the target. Let  be the angle between the line that links the robot to the
target and the axis x (see Figure 3). So:



   = arcsin (|yr – y’| / d)




Department Of EEE                             13                          MESCE Kuttipuram
                                                                          Seminar Report „04




The  can then be calculated as follows:



If  +  >  and  > 0 then  =  - ( - )

If  +  <=  and  >= 0 then b =  - ( + )

If  < 0 then  = - + ( + )



To determine the direction of the target relative to the robot, the difference between  and
 angles is calculated. The resulting direction is illustrated in Figure 4.




           Figure 4      Goal positioning




           Table 1     Goal positioning




Department Of EEE                             14                         MESCE Kuttipuram
                                                                          Seminar Report „04




                 CONTROLLER’S INPUTS AND OUTPUTS




There are 6 linguistic variables as inputs to the fuzzy control system:



     · The four sensor readings – right, left, ahead, behind



     · The angle between the robot and the target (β) in radians;



       ·The distance to the target, this calculated using the coordinates given by the
        simulator.




The output of the fuzzy system is represented by two linguistic variables, which are the
power to be applied in each step motor (left and right motors).




        Complementing these variables, a path memory mechanism is included as another
input variable, in order to prevent the robot from getting stuck by specific obstacle
formations. The path memory mechanism consists of a vector of 360 binary positions,
indexed by the angle β converted to degrees. A 0 (zero) value represents that the robot
tried to follow the direction indicated by that angle and failed (a collision occurred),
while a 1 (one) value means that the direction was never tried or didn‟t cause any
collision on the last n times it was tried (n is a parameter).




Department Of EEE                             15                          MESCE Kuttipuram
                                                                      Seminar Report „04




In the situation shown in Figure 5, if M_Path[100] = 1, the robot is allowed to go in that
direction, otherwise, when M_Path[100] = 0, the robot will be prevented from going in
that direction by the action of the path memory mechanism.




                   Figure 5 Path memory mechanism




Department Of EEE                          16                        MESCE Kuttipuram
                                                                       Seminar Report „04




                                     FUZZY SETS




The 6 linguistic variables described are divided in the following fuzzy sets:




        Motor Power (output variables) – divided into 5 fuzzy sets:

        Negative_High, Negative_Medium, Zero, Positive_Medium and Positive_High



        Angle  – divided into 5 fuzzy sets:

         Negative_Big, Negative_Medium, Zero, Positive_Medium and Positive_Big.



        Distance to the target – also divided into 4 fuzzy sets:

          Very_Far, Far, Near, and Very_Near..




Department Of EEE                            17                        MESCE Kuttipuram
                                                         Seminar Report „04




                        MEMBERSHIP FUNCTION

        Power




            Figure 6   Membership function of power

        Angle




           Figure 7 Membership function of power

        Distance




           Figure 8    Membership function of distance


Department Of EEE                       18               MESCE Kuttipuram
                                                Seminar Report „04




                                RULE SETS




                    Table 2   Fuzzy rule sets




Department Of EEE                        19     MESCE Kuttipuram
                                                                        Seminar Report „04




        Due to the high number of input variables in the control system (each with at
least 4 fuzzy sets), the rules set had to be simplified, in order to have a rule base of
reasonable size. This simplification was accomplished by eliminating rules with low or
even no probability to occur, and rules that cause the same effect in the robot movement.
The final rule base, presented in the table 2 , has been divided into three basic groups:



Straight movement, where the robot has either no obstacle in the target direction or the
obstacle is far;



Turn , where the robot has to turn in order to move to the target position; and



Detour, where the robot has to make a detour in order to avoid the obstacle detected.




Department Of EEE                            20                         MESCE Kuttipuram
                                                                      Seminar Report „04




                                      INFERENCE




The inference method used in this system was the MAX-MIN, where the minimum
operator is used as the implication method for each rule, and then all rules are composed
by the maximum operator of the resultant pertinence of each rule, in order to evaluate the
final output fuzzy set. The connective AND has been implemented using also the MIN
operator.




            As an example, suppose that the input sensor values result in the following
pertinence (μ (x)) for the input fuzzy sets:

μFAR(ahead)=0.9;

μVERY-FAR(ahead)=0.2;

μNEG-BIG(angle)=0.7;
μNEG-MEDIUM(angle)=0.1; and

M_Path [β’] =1.



       Then, the activation of the following two rules with the same fuzzy set in the
consequence can be evaluated as follows:




RULE 1: IF Ahead is far AND angle is negative_big AND M_Path[β’] THEN right
motor is positive_medium




Department Of EEE                              21                    MESCE Kuttipuram
                                                                    Seminar Report „04




RULE 2: IF Ahead is very_far AND angle is negative_medium AND M_Path[β’] THEN

         right motor is positive_medium



RULE 1 – Min (0,9,0,7, 1) = 0,7



RULE 2 – Min (0,2,0,1, 1) = 0,1



Composition of the rules – Max (0,7,0,1) = 0,7.



Therefore, the output fuzzy set positive_medium will be evaluated with pertinence 0.7.




Department Of EEE                          22                       MESCE Kuttipuram
                                                                       Seminar Report „04




                               DEFUZZYFICATION


 Two defuzzification methods canbe implemented: the weighted mean of maxima, and
the centroid, both has large application in control problems. Since the two methods
presented very similar performance, the weighted mean of maxima has been chosen as
the final method. In this method, the output value is given by the mean of the maximum
values of each output set weighted by the resultant pertinence of the inference process.

       As an example, suppose that the inference process resulted in the output fuzzy set
shown in bold in Figure 9




               Figure 9     Output Membership function



The crisp power value to be applied to the specific motor would be:

          P = ∑piui / ∑ui = ((0.5 x -8) + (0.7 x –2) + (0.2 x 0)) = - 3.8

                                      (0.5 + 0.7 + 0.2)

As the step motor only accepts integer values, the power applied is rounded to – 4.




Department Of EEE                           23                        MESCE Kuttipuram
                                                                      Seminar Report „04




                                ADVANTAGES



    Fuzzy based systems enable robots to deal with imprecise, ambiguous or
      uncertain information or situations.



    More effective that is this takes less number of steps than an algorithm based on
      crisp rule.



    Robots do not get struck in difficult situation, avoiding repeated paths.



    These can be small in size and can access area which is not safe for human life or
      where human can‟t access




Department Of EEE                            24                      MESCE Kuttipuram
                                                                   Seminar Report „04




                              DISADVANTAGES



    The design of the controller is very complex. It requires a skill designer with
      complete knowledge about the application domain of the controller.



    Robot detects the ditches only when it comes very near to it and hence requires a
      lot of steps to avoid ditches.




Department Of EEE                        25                       MESCE Kuttipuram
                                                                         Seminar Report „04




                                 APPLICATION


    Robotic monitoring of underground cables

              For many years, the main maintenance strategy for power cable system has
   been corrective maintenance that is there is no maintenance reaction until an
   unexpected failure. These lead to heavy loss in industries etc. so condition based
   maintenance is becoming a superior choose. Robotic monitoring is proved to be a
   viable solution to the maintenance of underground cables system. These robots use
   infrared sensors, the acoustic sensors and the fringing electric field sensors to identify
   the aging status and location of failure of the power cable system.



    Autonomous mine detecting robot

          De mining is a very risky operation when there is direct involvement of
      human. Also it is very slow process as there is no suitable mine detector available
      and it very costly. This autonomous mine detector robot provides cheap de
      mining, which reduce the human involvement as well as speed up de mining.



    Monitoring nuclear reactors



    Transport and surveillance



    Research and science works




Department Of EEE                           26                           MESCE Kuttipuram
                                                                    Seminar Report „04




                                CONCLUSION


Using of the fuzzy logic system in mobile robot increases it effectiveness. The robot
nearly takes 40% less number of step than an algorithm based on the crisp rules. In
addition due to the path memory mechanism, the robots do not get struck.



Due to these feature there are wide range of application in power system and de mining
where human risk is involved etc.




Department Of EEE                         27                       MESCE Kuttipuram
                                                                   Seminar Report „04




                               REFERENCES


    Nikos C. Tsourveloudis, Kimon P.Valavanis, and Timothy Hebert, “Autonomous
      Vehicle Navigation Utilizing Electrostatic Potential Fields and Fuzzy Logic”
      IEEE Transactions on robotic control and automation, vol.17, No. 4



    Hartmut Surmann, Jörg Huser and Liliane Peters “A Fuzzy System for Indoor
      Mobile Robot Navigation”, Proc. of the Fourth IEEE Int. Conf. on Fuzzy
      Systems, pp. 83 - 86, 20 – 24.03.1995, Yokohama, Japan.



    Marley Maria B. R. Vellasco ,”Mobile Robot Control Using Fuzzy Logic”
      Computer Science Department and Information Sciences Institute University of
      Southern California



    Ming Cao and Ernest Hall, “Fuzzy Logic Control for an Automated Guided

      Vehicle” Center for Robotics Research, University of Cincinnati




Department Of EEE                        28                       MESCE Kuttipuram
                                                        Seminar Report „04




                          LIST OF FIGURE


        Figure 1     Fuzzy controller Block Diagram          5

        Figure 2     Positions of sensors and motors         9

        Figure 3     Calculation of the angle between
                    robot and the target position           10

        Figure 4     Goal positioning                       11

        Figure 5     Path memory mechanism                  13

        Figure 6     Membership function of power           15

        Figure 7     Membership function of angle           15

        Figure 8     Membership function of distance        15

        Figure 9     Output Membership function             19




        LIST OF TABLES


        Tables 1    Goal positioning                        11

        Tables 2    Fuzzy rule sets                         16




Department Of EEE                       29              MESCE Kuttipuram

								
To top