Digital games design-Game AI

數位遊戲設計

DIGITAL GAMES DESIGN

Week14 21/May/2008 Instructor: I-Hsien Ting

AI for Games

 _{FSM (Finite State Machine)}  _{Fuzzy Logic}

 _{Decision Tree}

 _{Genetic Algorithm}

 _{Artificial Neural Networks}

The 2007 Game AI Award

-Half life 2 (顫慄時空 2)

 _{1. Valve has taken NPC interaction to the next-level}

with this game.

 _{2. The combine soldiers are appear very effective at}

coordinating to assault the player.

 The AI technology behind the squad behaviors has been maturing well since the original Half-Life, even if this episode didn’t show it off as much as previous games.  _{3. The game features an impressive variety of}

characters

 _{including zombies, robotic dogs and mechanical striders,}

etc.

 Each of them requires a custom AI, but they all tie into the story very well and work together to improve the player’s experience.

AI?? An AI Test of Half-Life

2

How to define Game AI

 Pac Man

 _{Game AI? Yes! No!}

The definition of game AI

 Paths Finding and Collision Detection

sometimes also have been included into game AI

 Broad definition of game AI

 _{Deterministic (定性)}  _{Weak AI}

 _{Narrow definition of game AI}

 _{Nondeterministic (非定性)}  _{Strong AI}

Weak AI and Strong AI

 _{Weak AI}

 Path finding

 A* Path finding

 Collision Detection

 Finite state machine (FSM;有限狀態機)

 If ~then rules

 Fuzzy Logic  _{Strong AI}

 Decision Tree, Neural Network, Genetic Algorithm, Probabilistic Method, Reinforcement Learning…

 Behavior Cloning

 Artificial Life

Artificial Life

Artificial Life

 Genetic Evolution

 _{http://www.generation5.org/jdk/demos/dictyAppl} et.html

Reinforcement Learning

Commercial Game AI

 What’s Different

 _{Normally more than one opponent: NPC teams}  _{Some NPC’s are opponents and others are there}

to support the player

 Rationality

 _{How rational is it for a weaker NPC to attack a}

stronger player head-on

 _{Believability is a major concern}

 _{NPC’s should challenge the player, but not be}

Visibility

 Additional intelligence adds complexity

 Only useful if it’s visible to the player

 _{Player should be able to explain AI behavior}

 Simpler methods are often preferred

 Goal: give the impression of intelligent behavior

AI Wish List

 AI should be intelligent, but flawed: make the AI as intelligent as possible and then dumb it down.

 _{The AI should have no blatant flaws:}

 _{No “golden path” to defeat the AI every time in the}

same way

 AI Should not look plain dumb or fail miserable in any situation

 _{AI should be efficient}

 Real-time reactions

 AI normally gets no more than 20% of the CPU time

 _{AI should be configurable by game designer or}

Dumbing Down

 Make NPC less accurate (e.g. in shotting)

 NPC with longer reaction times

 _{Engaging the player only one at a time}

 NPC making themselves more vulnerable by

moving around in exposed areas or changing positions often

Cheating

 Cheating is fine for the AI, but don’t get

caught. Often the easier and more efficient solution

 Sometimes, the player can be told (e.g. AI will build new units 25% faster)

 _{Initial position of NPCs is pre-designed (e.g.} traps, defense positions)

A* Path Finding

 Perform heuristic best-first search  Heuristic evaluation of node n

 _{Distance from initial state}

 _{Estimated distance to goal state}

 _{Admissible heuristic}

Collision Detection