19
Embracing Emergent Behavior with Goal- Based AI Kevin Dill [email protected]

Embracing Emergent Behavior with Goal-Based AI

  • Upload
    linh

  • View
    36

  • Download
    0

Embed Size (px)

DESCRIPTION

Embracing Emergent Behavior with Goal-Based AI . Kevin Dill [email protected]. Blue Fang Games. “To create compelling, emotionally engaging games, focused on the animal kingdom, that set the benchmark for broad-based family entertainment.”. The Plan. Motivation Kohan 2 Demo - PowerPoint PPT Presentation

Citation preview

  • Embracing Emergent Behavior with Goal-Based AI Kevin [email protected]

  • Blue Fang GamesTo create compelling, emotionally engaging games, focused on the animal kingdom, that set the benchmark for broad-based family entertainment.

  • The PlanMotivationKohan 2 DemoCommon AI TechniquesGoal-based AIBoxy-dog DemoQuestions

  • The PlanWhat I Want:Cunning OpponentsResponsive AlliesReplayabilityOrganic Credibility Emotional AppealCompelling PerformancesOr, more generally:Situational AwarenessEmergent Behavior

  • Motivation, cont.Declarative AI (vs. Procedural AI)Wikipedia: A program is declarative if it describes what something is like, rather than how to create it.Developers role: not to tell the AI what to do, but to tell the AI how to decide for itself.

  • Techniques Scripted AIAI Follows a script, much like a movieScript specifies exactly what will happen, when it will happen, where it will happenAdvantages:Gives designers explicit control Easy to writeEasy to balanceDisadvantagesRepetitive and predictable

  • Techniques State MachinesA state is defined for every high-level action the AI can takeTransitions are defined to move from one state to another stateEach state has custom code to control the character while in that state

  • First Person Shooter ExampleGet AmmoGet HealthRun AwaySearch For EnemyAttackEnemy FoundEnemy LostLow on AmmoLow on HealthGot Away, Low on HealthGot HealthGot AmmoGot Away, Low on Ammo

  • Techniques State MachinesAdvantages:Easy to conceptualizeEasy to implementMiddleware tools existDisadvantages:Complexity scales poorly as AI growsRampant duplication of codeCan only do one thing at a time

  • Techniques Goal Based AIA goal is defined for every high-level action that the AI can takeEach goal is assigned a priority:Base priority (evaluate current situation)Allocate resourcesOptimize resources and calculate final priorityHighest priority goal(s) are executed

  • Goal TypesSome common RTS goals:Attack Construct BuildingDefend Repair BuildingExplore Upgrade BuildingRecruit Unit Sell BuildingRecover Health Give Resources To AllyResearch

  • Techniques Goal Based AIAdvantagesFlexible and powerfulComplexity scales fairly well as AI growsCan execute more than one goal at a timeEmergent behaviorDisadvantages:Priority calculations (Bucket of Floats)Emergent behavior

  • Goal Selection BasicsWWHD (What Would a Human Do?)Write down a list of the factors you would consider if you were playing the gameSome common tricks:Bonuses Repeat PenaltiesMultipliers One-Time BonusesExponents Actor CategoriesFuzzy Factors Templates

  • Example: Attack GoalBase priority:Actor value for the most desirable building or unit in the area (taking into account enemys economy if appropriate)Travel DistanceIntervening targets (attack goals only)

  • Attack Goal, cont.Priority Curve:Target CV ratioMin CV to assignLow CV exponent Max advantageMax multiplier Other considerationsCombined armsWall strengthAdjusting builder CV* CV = Combat Value

  • Loose EndsGoal CommitmentIF: you cant afford something nowBUT: you will be able to soonTHEN: dont perform lower priority goals that use the same resourceGoal InertiaBonus to keep doing what youre doingLocked GoalsDisable all other goals until current goal is complete

  • Jack the Terrier