Docstoc

蟻群算法應用在足球機器人的避障路徑上

Document Sample
蟻群算法應用在足球機器人的避障路徑上 Powered By Docstoc
					Southern Taiwan University




         PSO-based Fuzzy Controller
         Design for Robot Soccer
       Department of Electrical Engineering, Southern Taiwan University,
       Tainan, R.O.C


        Juing-Shian Chiou, Chi-Jo Wang, Shih-Wen Cheng,
        Kuo-Yang Wang, and Yu-Chia Hu




                                           1
Southern Taiwan University




     Outline
        Abstract
        Introduction
        Motion Fuzzy Controller Structure
        Particle Swarm Optimization algorithm
        Simulation Results
        Conclusion



                              2
Southern Taiwan University




      Abstract
     This paper will use the algorithm based on fuzzy to combine with
      particle swarm algorithm, applying to the mobile robot’s obstacle
      avoidance, determine the fuzzy algorithm and Particle Swarm
      Optimization (PSO) to design the optimal route and speed.

     In this paper, we will use this algorithm in five-versus-five
      simulation platform, it knows whether the combination of these
      algorithms can be quickly and accurately to achieve our objectives.




                                        3
Southern Taiwan University




      Introduction(1/3)
     we propose the application of this algorithm - fuzzy particle swarm
      algorithm, the advantage of PSO, convergence time is quicker than
      others and easy to modify, the above features are very special in the
      algorithm.

     In this experimental platform, we choose the robot Football
      Association 5-vs-5 simulation platform in Figure 1




                             Figure 1. The Five-versus-Five simulation platform

                                         4
Southern Taiwan University




      Introduction(2/3)
     Using to cluster features for the particle swarm, to find out how to
      avoid obstacles in the move, at the same time moving towards the
      destination path planning, and this focus on how to quickly take the
      lead particles are individual optimal solution, also obtained group
      optimal solution, show in Figure 2.




                                         5
Southern Taiwan University




     Introduction(3/3)
                          Set of fuzzy rule.




               Fuzzy controller of robot wheels
                           Speed.




                   Evaluate each particle's fitness
                             function.



              Records of individual particles and groups
                       of the best memories.




                       Update the particle
                      position and velocity.




              No
                     Terminating condition.
                                                           Figure 2. System structure.
                                  Yes

                               END                         6
Southern Taiwan University

   Motion Fuzzy Controller
   Structure(1/7)
     In this part, we start design the fuzzy logic controller aimed at
      producing the velocities of the robot right and left wheel. We set
      two input parameters of the fuzzy logic controller are distance d
      and angle  .


     The former d is the distance between the robot and the goal. The
      latter  is the direction of with on the straight line path to the
      goal. Both are shown in Figure 3.




                                      Figure 3. the relation of d and 

                                         7
Southern Taiwan University

      Motion Fuzzy Controller
      Structure(2/7)
     We set the values of variable e1 , e2 , e3 , e4 , v1 , v2 , y1 , y 2 and design two
      fuzzy controllers to control the velocity of the right and left wheels to
      move the robot.

     The fuzzy rules on which were based these fuzzy controllers are
      described in tables 1 and 2, and can be described according to the
      following equations:
           Ry1  j1 , j2  : IF    e1   is   A1, j1    And   e2   is   A 2, j2 

      Then
             y1 is y1 j1 , j2   j1, j2 3, 2, 1,0,1,2,3            (1)
           Ry2  j3 , j4  : IF e3 is  3, j1  And e4 is  4, j2 
                                         A                     A
      Then
             y2 is y2 j3 , j4    j3 , j4 3, 2, 1,0,1,2,3          (2)
                                                         8
Southern Taiwan University

     Motion Fuzzy Controller
     Structure(3/7)
  Table 1. Fuzzy rule base of the left-   Table 2. Fuzzy rule base of the right-
     wheel velocity fuzzy controller                wheel velocity controller




                                           9
Southern Taiwan University

      Motion Fuzzy Controller
      Structure(4/7)
     The following term sets were used to describe the fuzzy sets of each
      input and output fuzzy variables:
               T  ei    NB, NM , NS , Z , PS , PM , PB , i  1,2,3,4

                                                                                            
                    Ai ,3 , Ai ,2 , Ai ,1 , Ai ,0 , Ai ,1 , Ai ,2  , Ai ,3 ,
                                                                                                     (3)


               T  ym    NB, NM , NS , Z , PS , PM , PB , m  1,2

                     
                    y m,3 , y m,2 , y m,1 , y m,0 , y m,1 , y m,2  , y m,3 ,      (4)




                                                                     10
Southern Taiwan University

      Motion Fuzzy Controller
      Structure(5/7)
     As show in figure 4, the triangle membership function and the
      singleton membership function are used to describe the fuzzy sets
      of input variables and output variables.


                                                                                    
      NB   NM     NS   Z     PS   PM   PB                  NB     NM     NS     Z        PS    PM    PB




      0    10     20   30    40   50   60   (inch)
                                                          90   60   30   0        30   60   90   

                            (a)                                                (b)
                Figure 4. Membership function: (a) the fuzzy sets for ei ;
                                               (b) the fuzzy sets for ym .

                                                     11
 Southern Taiwan University

             Motion Fuzzy Controller
             Structure(6/7)
                Based on the weighted average method, the final output of these
                 fuzzy controllers can be described by means of equation (5) and (6)

                 3         3                                                                                    3           3
y1           w
           j1 3 j2 3
                                       j1 , j2 
                                                          y1 j1 , j2                 (5)     y2            w
                                                                                                           j3 3 j4 3
                                                                                                                                         j3 , j4 
                                                                                                                                                            y1 j3 , j4         (6)



                Where w j1 , j2  and w j3 , j4  were determined according to Equations
                 (7) and (8).


w j1 , j2  
                                 
                         min u A1, j   e1  , u A2, j   e2 
                                       1                    2
                                                                                       (7)    w j3 , j4  
                                                                                                                                
                                                                                                                      min u A3, j   e3  , u A 4, j   e4 
                                                                                                                                     3                       4
                                                                                                                                                                                (8)
                   min  u  e  , u                                       e2                                min  u                            e3  , u A   e4  
                     3     3                                                                                     3      3

                                           A1, j    1         A 2, j                                                                  A 3, j               4, j4
                                                 1                     2                                                                         3
                 j1 3 j2 3                                                                                  j1 3 j2 3




                                                                                              12
Southern Taiwan University

      Motion Fuzzy Controller
      Structure(7/7)
     When the input data of e1 , e2 , e3 and e4 are given, y1 and y2 can be
      determined by using Equations (5) and (6) Thus, the left-wheel velocity
      vl and the right-wheel velocity vr can be obtained.




                                        13
Southern Taiwan University

     Particle Swarm Optimization
     algorithm (1/3)
        Initially the group is based on the flock-based mobile way, there are
         based on particles, the first, particles will be randomly distributed in
         space, each particle has own optimal solution, also to the group's
         information to determine the optimal location of the next movement
         and speed, the optimal by different individuals, repeat implementation
         to find the overall optimization, after iterative calculation method, to
         achieve the optimization goal.

        In the particle swarm system in the simultaneous existence of
         individual optimal value pbesti and the group optimal value gbesti ,
         it will use the robot to avoid obstacles function, the schematic diagram
         of pbesti and gbesti below.


                                         14
Southern Taiwan University

     Particle Swarm Optimization
     algorithm (2/3)
         v  wv  c  rand ()  ( gbest  s ) 
          k 1

          i
                       k

                       i       1
                                                       k*

                                                       i
                                                               k

                                                               i   (9)
                      c  rand ()  ( pbest  s )
                                              #
                                          k                k

                       2                  i                i




                                                                   (10)
         s    k 1

              i
                      s vk

                           i
                                   k 1

                                   i



        According to the above function, determining the velocity and
        position, the maximum speed limit vmax for each particle, and
        the maximum distance limit smax ,When the speed limit and
        greater limit than distance, The speed and distance will be
        defined as vmax or smax 。

                                                  15
Southern Taiwan University

     Particle Swarm Optimization
     algorithm (3/3)




             Figure 5. Particle velocity and position graph


                                         16
Southern Taiwan University




      Simulation Results(1/2)
     We use the particle swarm algorithm to simulate the path and the
      avoidance function.




                  (a)                                   (b)
      Figure 6(a)(b). Use the PSO to modify the fuzzy rule, the robot to
      achieve faster
                                        17
Southern Taiwan University




     Simulation Results(2/2)




                     (a)                                (b)

      Figure 7(a)(b). While planning a path ahead to avoid obstacles in
      the movement
                                        18
Southern Taiwan University




      Conclusion
     In this experiment, we use particle swarm algorithm to avoid
      obstacles, at the same time toward the destination, and through the
      particle swarm faster convergence to obtain the optimal solution, can
      achieve the path planning objects quickly, at the same time as change
      with the environment, and immediately change its pre-determined
      parameters, PSO is easy to change, the platform is also very easy to
      operate.




                                       19
Southern Taiwan University




           Thanks for your attention !




                             20

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:17
posted:3/31/2013
language:Unknown
pages:20