Digital games design-Game AI II

數位遊戲設計

DIGITAL GAMES DESIGN

Week15 28/May/2008 Instructor: I-Hsien Ting

AI for Games Part II



Finite State Machine



Behavioral Cloning



_{Fuzzy Logic}



Probability

Finite State Machine (有限狀

態機)



Intelligent Agents

Finite State Machine



Decopose an agent’s behavior into easily

manageable chunks (“states”)



Transitions between the states are activated

by environment or agent-internal events

Why FSM??



Quick and simple to implement



Easy to debug



_{Computationally efficient}



Intuitive

Famous Example: Who’s Who of

FSMs



_{Pac-Man ghost}

 _{Evade and Chase states, Evade state same for all}

ghosts, Chase state individual to each ghost



_Quake

 _{States include FindHealth, SeekCover, RunAway,}

Weapons also are implemented as FSMs: Move, TouchObject, Die



FIFA

 Strike, Dribble, ChaseBall, MarkPlayer



_Warcraft

FSM Implementation

FSM Implementation



Problems of quick and dirty approach

 _{As more state are added, program becomes more}

difficult to understand, and debug

 _{Inflexible: difficult to extend beyond original}

purpose

 _{Not possible to treat entry into and exit out of}

state in a structured manner (e.g. scream when starting an attach)

 _{FSM tend to grow quickly and become too}

FSM Example



Ants

 _{http://im.nuk.edu.tw/~iting/course/im_game_200} 8a/Ants.rar Food Die Go Home Thirsty 找到毒物 找到食物 _找到水 找到毒物 找到毒物 找到巢穴

Behavioral Cloning



What’s it all about?

 _{Reconstruction of skill from an operators’ control}

(or action) traces by means of machine learning

 _Examples:

 Flying a plane

 Operating a crane

Behavioral Cloning



Why?

 _{Control Skills are often acquired through}

experience and are sub-cognitive

 _{Thus, it is hard to reconstruct and implement in a}

computer program

Behavior Cloning



Behavior Trace

 _{Trace is set of examples of correct control decision}  _{Each example consists of a pair}

 _{Time delay between state and action to account}

for time needed by operator for state recognition and physical action

Behavior Cloning



Machine Learning

 _{Action(Time)=f(State(Time-Delay))}  _{Goal: learn function f from examples}

Decision Tree Learning

Decision Tree Learning

Fuzzy Logic



Fuzzy Logic is an extension of classical logic

that is based on the idea of a fuzzy set.

Distance Value At 0.00..0.15 Very Close 0.10..0.35 Close 0.30..0.70 Pretty Far 0.60..0.90 Very Far 0.85..1.00

Fuzzy Logic

Distance Value At 0.00..0.15 Very Close 0.10..0.35 Close 0.30..0.70 Pretty Far 0.60..0.90 Very Far 0.85..1.00

Fuzzy Logic in Game



Control

 _{Train, Air Condition, Robot}

 E.g. Turning Smoothly



Threat Assessment

Control

Fuzzy Set 歸屬程度很左邊 0.0左邊 0.0 前方 0.0 右邊 0.47 很右邊 0.3 mFarLeft = FuzzyReverseGrade (33, -40, -50) mLeft = FuzzyTrapezoid (33, -40, -25, -15, 0) mAhead = FuzzyTriangle (33, -10, -10) mRight = FuzzyTrapezoid (33,0,15,25,40) mFarRight = FuzzyGrade (30, 30,40);

Threat Assessment

mClose = FuzzyTriangle(25, -30, 0, 30); mMedium = FuzzyTrapezoid(25, 10, 30, 50, 70); mFar = FuzzyGrade(25, 50, 70); mTiny = FuzzyTriangle(8, -10, 0, 10); mSmall = FuzzyTrapezoid(8, 2.5, 10, 15, 20); mModerate = FuzzyTrapezoid(8, 15, 20, 25, 30); mLarge = FuzzyGrade(8, 25, 30); Fuzzy Set 歸屬程度 逼近 0.17 附近 0.75 遠 0.0 零星 0.2 少量 0.73 中等 0.0 大型 0.0

Rule Matrix

逼近 附近 遠 零星 中等 低等 低等 少量 高度 低等 低等 中等 高度 中等 低等 大型 高度 高度 中等

Random and Probability



25 % Left



25% Right



_{50% Back}



Random (0-99)

 _0~24Left  _25~74Back  _{75~99Right}

Status Transfer