Fuzzy Behavior-Based Control of Mobile Robots

Prahlad Vadakkepat, Ooi Chia Miin, Xiao Peng, and Tong Heng Lee

IEEE TRANSACTIONS ON FUZZY SYSTEMS

Introduction

Behavior-based architectures help to decompose control systems into subsystems with task achieving behaviors.

fuzzy logic has been employed in implementing individual robot behaviors and actions, in particular for shooting and obstacle avoidance

The Behavior-Based Architecture for Robot Soccer System

Communication:RF The whole soccer team 分為單一行為 & 集體行為模式 At the top of the hierarchy isthe overall team

behavior,which istermed asthe team strategy. Layer 3:These behaviors are further decomp

osed into simpler actions such as go-position and go-angle.

Obstacle avoidance is explicitly fulfilled by independent behaviors, such as shun-robots and avoid-wall, while other behaviors need not take it into consideration.

其他的報告是認為直接包成一個行為其中之一的因素

Fuzzy Rule-Base Coordination:

Fuzzy Rule-Base Coordination:

It’s a topd-down decision making approach Two methods: center of area(CoA),max criter

ion CoA:The CoA method is used to “merge” co

mpatible behaviors and actions. Max criterion:The max criterion activatesonly

one behavior at any time

Activity and Action Contribution:

Activity and Action Contribution:

It’s bottom-up approach Each fuzzy action provides a pair of activity

and action values. Action represents the output of a fuzzy action, and in this case, is the pair of wheel speeds. Activity represents the degree of contribution a fuzzy action has on the overall behavior.

Activity and Action Contribution:

The CoA method takes the average of the action values recommended by the fuzzy actions, weighted by the activity values.

The max criterion activates the fuzzy action with the greatest activity value. The application of either method generally depends on whether the actions are mutually exclusive or not.

Method of Coordination:

behavior coordination is also based on fuzzy rule-base coordination.

守門員是單純的 if –else-if 構成的 參數的雙方取決問題 Ex. a role that contains the avoid-wall behavior can n

ot activate that behavior based on the fuzzy rule. The avoid-wall behavior recommends itself when it is necessary.

Implemententation

The fuzzy behavior-based architecture resulted in a distributed fuzzy system with smaller subsystems.

Ex. the get-ball-at-angle action is described. This action is implemented based on the go-position-at-angle action, with the ball asthe target position.

輸入參考變數 : the distance between the ball and the robot the speed of the ball the direction of the ball relative to the robot’s heading angle. 遠離 ?or 正朝著 robot 滾來 ?

Reactive & deliberative

Reactive behavior is defined as that displayed in direct response to environmental stimulation.

Ex. avoid-wall, shun-robots, and frustration.

Deliberative

deliberative behavior is displayed not purely in response to environmental situations

Ex. shooting, chasing, passing, wandering, guarding,sweeping, blocking and tracking.

特性 :very flexible

結論

將行為模式階層化會很有效率 複雜的行為模式 ( 射門 , 阻擋 ), 由 sub fuzzy s

ystem 逐步建構起來 , 可以有效率分析 最基本行為的優先權 ( 不撞牆 ) 不能跟高等行

為混在一起 .