Artificial Intelligence For GAMES

895 Pages·2009·4.09 MB·English

ARTIFICIAL
INTELLIGENCE
FOR GAMES
Second Edition
IANMILLINGTONand JOHNFUNGE
AMSTERDAM•BOSTON•HEIDELBERG•LONDON
NEWYORK•OXFORD•PARIS•SANDIEGO
SANFRANCISCO•SINGAPORE•SYDNEY•TOKYO
MorganKaufmannPublishersisanimprintofElsevier
MorganKaufmannPublishersisanimprintofElsevier.
30CorporateDrive,Suite400,Burlington,MA01803,USA
Thisbookisprintedonacid-freepaper.
Copyright©2009byElsevierInc.Allrightsreserved.
Designationsusedbycompaniestodistinguishtheirproductsareoftenclaimedastrademarksorregisteredtrademarks.Inall
instancesinwhichMorganKaufmannPublishersisawareofaclaim,theproductnamesappearininitialcapitalorallcapital
letters.Alltrademarksthatappearorareotherwisereferredtointhisworkbelongtotheirrespectiveowners.NeitherMorgan
KaufmannPublishersnortheauthorsandothercontributorsofthisworkhaveanyrelationshiporaffiliationwithsuch
trademarkownersnordosuchtrademarkownersconfirm,endorseorapprovethecontentsofthiswork.Readers,however,
shouldcontacttheappropriatecompaniesformoreinformationregardingtrademarksandanyrelatedregistrations.
Nopartofthispublicationmaybereproduced,storedinaretrievalsystem,ortransmittedinanyformorbyany
means—electronic,mechanical,photocopying,scanning,orotherwise—withoutpriorwrittenpermissionofthepublisher.
PermissionsmaybesoughtdirectlyfromElsevier’sScience&TechnologyRightsDepartmentinOxford,UK:phone:(+44)1865
843830,fax:[phone],E-mail:[e-mail]questonlineviatheElsevier
homepage(http://elsevier.com),byselecting“Support&Contact”then“CopyrightandPermission”andthen“Obtaining
Permissions.”
LibraryofCongressCataloging-in-PublicationData
Millington,Ian.
Artificialintelligenceforgames/IanMillington,JohnFunge.–2nded.
p.cm.
Includesindex.
ISBN978-0-12-374731-0(hardcover:alk.paper)
1.Computergames–Programming.2.Computeranimation.3.Artificialintelligence.
I.Funge,JohnDavid,1968-II.Title.
QA76.76.C672M5492009
006.3–dc22
2009016733
BritishLibraryCataloguing-in-PublicationData
AcataloguerecordforthisbookisavailablefromtheBritishLibrary.
ISBN:978-0-12-374731-0
ForinformationonallMorganKaufmannpublications
visitourWebsiteatwww.mkp.comorwww.elsevierdirect.com
Typesetby:diacriTech,India
PrintedintheUnitedStatesofAmerica
09 10 11 12 13 5 4 3 2 1
For Conor – I.M.
For Xiaoyuan – J.F.
About the Authors
IanMillington isapartnerof IcosagonLtd.(www.icosagon.com),aconsultingcompanydevel-
opingnext-generationAItechnologiesforentertainment,modeling,andsimulation.Previously
he founded Mindlathe Ltd.,the largest specialist AI middleware company in computer games,
working on a huge range of game genres and technologies. He has a long background in AI,
includingPhDresearchincomplexitytheoryandnaturalcomputing.Hehaspublishedacademic
andprofessionalpapersandarticlesontopicsrangingfrompaleontologytohypertext.
JohnFunge(www.jfunge.com)recentlyjoinedNetflixtostartandleadthenewGamePlatforms
group.Previously,Johnco-foundedAiLiveandspentnearlytenyearshelpingtocreateasuccessful
companythatisnowwellknownforitspioneeringmachinelearningtechnologyforgames.AiLive
co-createdtheWiiMotionPlushardwareandhasestablisheditsLiveMoveproductsastheindustry
standardforautomaticmotionrecognition.AtAiLiveJohnalsoworkedextensivelyonLiveAI,a
real-timebehaviorcaptureproductthatisbeingusedbytheformerleadgamedesignerofGuitar
HeroandRockBandtocreateanewgenreofgame.JohnisalsoanAssistantAdjunctProfessor
attheUniversityof California,SantaCruz(UCSC)whereheteachesaGameAIcoursethathe
proposed,designedanddeveloped.JohnhasaPhDfromtheUniversityofTorontoandanMSc
from the University of Oxford. He holds several patents, is the author of numerous technical
papers,andwrotetwopreviousbooksonGameAI.
iv
Contents
About the Authors
iv
Acknowledgments
xix
Preface
xxi
About the Website
xxiii
Part I
AI and Games 1
Chapter
1
Introduction
3
What Is AI?
1.1 4
1.1.1 AcademicAI 5
1.1.2 GameAI 7
Model of Game AI
1.2 8
1.2.1 Movement 9
1.2.2 DecisionMaking 10
1.2.3 Strategy 10
1.2.4 Infrastructure 11
1.2.5 Agent-BasedAI 11
1.2.6 IntheBook 12
Algorithms, Data Structures, and
1.3
Representations
12
1.3.1 Algorithms 12
1.3.2 Representations 15
v
vi
Contents
On the Website
1.4 16
1.4.1 Programs 16
1.4.2 Libraries 17
Layout of the Book
1.5 18
Chapter
2
Game AI
19
The Complexity Fallacy
2.1 19
2.1.1 WhenSimpleThingsLookGood 19
2.1.2 WhenComplexThingsLookBad 20
2.1.3 ThePerceptionWindow 21
2.1.4 ChangesofBehavior 21
The Kind of AI in Games
2.2 22
2.2.1 Hacks 22
2.2.2 Heuristics 23
2.2.3 Algorithms 24
Speed and Memory
2.3 25
2.3.1 ProcessorIssues 25
2.3.2 MemoryConcerns 28
2.3.3 PCConstraints 29
2.3.4 ConsoleConstraints 29
The AI Engine
2.4 31
2.4.1 StructureofanAIEngine 32
2.4.2 ToolchainConcerns 33
2.4.3 PuttingItAllTogether 34
Part II
Techniques 37
Chapter
3
Movement
39
The Basics of Movement Algorithms
3.1 40
3.1.1 Two-DimensionalMovement 41
3.1.2 Statics 42
3.1.3 Kinematics 45
Kinematic Movement Algorithms
3.2 49
3.2.1 Seek 49
vii
Contents
3.2.2 Wandering 53
3.2.3 OntheWebsite 55
Steering Behaviors
3.3 55
3.3.1 SteeringBasics 55
3.3.2 VariableMatching 56
3.3.3 SeekandFlee 56
3.3.4 Arrive 59
3.3.5 Align 62
3.3.6 VelocityMatching 66
3.3.7 DelegatedBehaviors 67
3.3.8 PursueandEvade 68
3.3.9 Face 71
3.3.10 LookingWhereYou’reGoing 72
3.3.11 Wander 73
3.3.12 PathFollowing 76
3.3.13 Separation 82
3.3.14 CollisionAvoidance 84
3.3.15 ObstacleandWallAvoidance 90
3.3.16 Summary 95
Combining Steering Behaviors
3.4 95
3.4.1 BlendingandArbitration 96
3.4.2 WeightedBlending 96
3.4.3 Priorities 103
3.4.4 CooperativeArbitration 107
3.4.5 SteeringPipeline 108
Predicting Physics
3.5 120
3.5.1 AimingandShooting 121
3.5.2 ProjectileTrajectory 121
3.5.3 TheFiringSolution 123
3.5.4 ProjectileswithDrag 126
3.5.5 IterativeTargeting 128
Jumping
3.6 134
3.6.1 JumpPoints 135
3.6.2 LandingPads 138
3.6.3 HoleFillers 143
Coordinated Movement
3.7 144
3.7.1 FixedFormations 144
3.7.2 ScalableFormations 146
3.7.3 EmergentFormations 146
3.7.4 Two-LevelFormationSteering 147
3.7.5 Implementation 151
viii
Contents
3.7.6 ExtendingtoMorethanTwoLevels 157
3.7.7 SlotRolesandBetterAssignment 159
3.7.8 SlotAssignment 162
3.7.9 DynamicSlotsandPlays 166
3.7.10 TacticalMovement 168
Motor Control
3.8 171
3.8.1 OutputFiltering 172
3.8.2 Capability-SensitiveSteering 174
3.8.3 CommonActuationProperties 175
Movement in the Third Dimension
3.9 178
3.9.1 RotationinThreeDimensions 178
3.9.2 ConvertingSteeringBehaviorstoThreeDimensions 180
3.9.3 Align 180
3.9.4 AligntoVector 181
3.9.5 Face 183
3.9.6 LookWhereYou’reGoing 186
3.9.7 Wander 186
3.9.8 FakingRotationAxes 188
Exercises
192
Chapter
4
Pathfinding
197
The Pathfinding Graph
4.1 198
4.1.1 Graphs 198
4.1.2 WeightedGraphs 199
4.1.3 DirectedWeightedGraphs 202
4.1.4 Terminology 203
4.1.5 Representation 203
Dijkstra
4.2 204
4.2.1 TheProblem 205
4.2.2 TheAlgorithm 206
4.2.3 Pseudo-Code 210
4.2.4 DataStructuresandInterfaces 212
4.2.5 PerformanceofDijkstra 214
4.2.6 Weaknesses 214
A*
4.3 215
4.3.1 TheProblem 216
4.3.2 TheAlgorithm 216
4.3.3 Pseudo-Code 220
4.3.4 DataStructuresandInterfaces 223
ix
Contents
4.3.5 ImplementationNotes 228
4.3.6 AlgorithmPerformance 228
4.3.7 NodeArrayA* 229
4.3.8 ChoosingaHeuristic 231
World Representations
4.4 237
4.4.1 TileGraphs 239
4.4.2 DirichletDomains 241
4.4.3 PointsofVisibility 244
4.4.4 NavigationMeshes 246
4.4.5 Non-TranslationalProblems 251
4.4.6 CostFunctions 251
4.4.7 PathSmoothing 251
Improving on A*
4.5 255
Hierarchical Pathfinding
4.6 255
4.6.1 TheHierarchicalPathfindingGraph 256
4.6.2 PathfindingontheHierarchicalGraph 259
4.6.3 HierarchicalPathfindingonExclusions 262
4.6.4 StrangeEffectsofHierarchiesonPathfinding 263
4.6.5 InstancedGeometry 265
Other Ideas in Pathfinding
4.7 271
4.7.1 OpenGoalPathfinding 272
4.7.2 DynamicPathfinding 272
4.7.3 OtherKindsofInformationReuse 273
4.7.4 LowMemoryAlgorithms 273
4.7.5 InterruptiblePathfinding 274
4.7.6 PoolingPlanners 275
Continuous Time Pathfinding
4.8 276
4.8.1 TheProblem 276
4.8.2 TheAlgorithm 277
4.8.3 ImplementationNotes 281
4.8.4 Performance 281
4.8.5 Weaknesses 282
Movement Planning
4.9 282
4.9.1 Animations 282
4.9.2 MovementPlanning 283
4.9.3 Example 286
4.9.4 Footfalls 287
Exercises
288