towards conscious-like behavior in computer game characters
DESCRIPTION
The main sources of inspiration for the design of more engaging synthetic characters are existing psychological models of human cognition. Usually, these models, and the associated Artificial Intelligence (AI) techniques, are based on partial aspects of the real complex systems involved in the generation of human-like behavior. Emotions, planning, learning, user modeling, set shifting, and attention mechanisms are some remarkable examples of features typically considered in isolation within classical AI control models. Artificial cognitive architectures aim at integrating many of these aspects together into effective control systems. However, the design of this sort of architectures is not straightforward. In this paper, we argue that current research efforts in the young field of Artificial Consciousness (AC) could contribute to tackle complexity and provide a useful framework for the design of more appealing synthetic characters. This hypothesis is illustrated with the application of a novel consciousness-based cognitive architecture to the development of a First Person Shooter video game character.TRANSCRIPT
Towards Conscious-like Behavior in Computer Game
CharactersRaúl Arrabales, Agapito Ledezma and Araceli Sanchis
Charactersg p
Computer Science DepartmentCarlos III University of Madrid
htt //C i R b t /R lhttp://Conscious-Robots.com/Raul
IEEE Symposium on Computational Intelligence and Games 2009IEEE Symposium on Computational Intelligence and Games 2009Politecnico di Milano, Milano, 9th September 2009
Contents
IntroductionMotivation and ObjectivesState of the ArtState of the ArtConsciousness and GamesExperimental EnvironmentExperimental EnvironmentProposed ArchitectureEvaluationConclusions
www.Conscious-Robots.com
Conclusions2
Introduction (I)Objective:
Design mo e appealing s nthetic cha acte sDesign more appealing synthetic characters.More engaging, more real.H lik (C i lik b h i )Human-like (Conscious-like behaviors).
Problem:Many cognitive skills involved.a y cog t e s s o edHow to integrate them effectively?
www.Conscious-Robots.com 3
Introduction (II)Character design challenges:
E ti l i l i tEmotions, planning, learning, opponent modeling, attention, set shifting, etc.How can they be integrated effectively?
Our hypothesis:Use a cognitive architecture based on a model of consciousness.
www.Conscious-Robots.com 4
Motivation and ObjectivesUltimate goal:
Cha acte s able to pass the (adapted) T ing testCharacters able to pass the (adapted) Turing test.
Why?State of the art video gamesg
Complex virtual environments.Elaborated scripts.Elaborated scripts.Require more realistic behaviors.
www.Conscious-Robots.com 5
State of the artCurrent AI characters cannot match human-like behaviorlike behavior.
E.g. see last year 2K Botprize contest results.
Usually, playing with other humans is more realistic and engaging that playing with AI bots.
Apply Machine Consciousness to games to improve believability?
www.Conscious-Robots.com
improve believability?6
Consciousness and Games (I)Embodiment and situatedness are claimed to be critical for the production of consciousness.p
We argue that consciousness-based softwareWe argue that consciousness based software agents can also be embodied and situated.
Software sensors and actuatorsSoftware sensors and actuators.Causal interaction with the world.
Virtual World (virtually situated).Virtual World (virtually situated).Real World (interaction with human players).
www.Conscious-Robots.com 7
Consciousness and Games (II)Games versus Physical Robots
Robots possess physical bodyRobots possess physical body.Robots interact directly with physical world.Robots have to deal with noise and uncertaintyRobots have to deal with noise and uncertainty.
Th t b b fitThese aspect can be seen as benefits or drawbacks
d l f f h h l l lGames are ideal for focusing on high-level control.
www.Conscious-Robots.com 8
Experimental EnvironmentFPS: Unreal Tournament 2004
Behavior Controller
B i B h i
Bot
AnimationModule
Movement Module
Basic BehaviorsSensory DataProcessing
ModuleModule
SensoryInputs Atomic Actions
BodyState
Virtual World
www.Conscious-Robots.com 9
Proposed Architecture (I)Context: Machine Consciousness Research.
Specifically: Design of machines showing conscious-like p y g gbehavior.
A C i (A C i ) A hAccess Consciousness (A-Consciousness) Approach:Only a small selection of contents gain access to a unique sequential thread for explicit processing and volition.q p p g
Computational model based on:Global Workspace Theory (Baars, 1988).Multiple Draft Model (Dennett, 1991).
www.Conscious-Robots.com
Proposed Architecture (II)Computational model of consciousness:
L b f i li dLarge number of specialized processors.
Competing and collaborating to access a global workspace (working memory).g p ( g y)
Where coherent information patterns areWhere coherent information patterns are selected.
www.Conscious-Robots.com
Proposed Architecture (III)Contexts
Working Memory
Focus of Attention
SpecializedProcessors Interim
Coalition
www.Conscious-Robots.com
Proposed Architecture (IV)CERA-CRANIUM Architecture
Implements the computational model of AImplements the computational model of A-Consciousness.
CERA is a layered control architecture.CRANIUM is a runtime component for the management of specialized processors.g p p
www.Conscious-Robots.com
Proposed Architecture (V)CERA-CRANIUM Perception Flow
Agent’s B d
CERA WorldBody
S AccessibleSensor Physical Layer
Sensors
Actuators
AccessibleenvironmentReachable
environment
Services
Motor Services
Mission‐specific
Core LayerServices
www.Conscious-Robots.com
Proposed Architecture (VI)CRANIUM Workspace (at CERA Physical layer)
CRANIUM WorkspaceSingle ComplexWorkspaceSingle
PerceptsComplex Percepts
…Percept
AggregatorsAggregators
www.Conscious-Robots.com
Proposed Architecture (VII)CRANIUM Workspace Modulation
Modulation Commands
Workspace WorkspaceCore Layer
Single Percepts
Complex Percepts
Mission Percepts
… …Physical Layer Mission‐Specific Layer
www.Conscious-Robots.com
Proposed Architecture (VIII)CRANIUM Workspace Modulation
UT2004 ServerCERA
UT2004 Viewer GameBots
CERA CRANIUM
UT2004 Viewer2004
UT2004 Client
UT2004 Client
CCR / DSS Runtime
Translated Pogamut 2
CERA‐CRANIUM UT2004 Client
(Human player)
UT2004 Client
(Spectator).NET Framework
www.Conscious-Robots.com
Proposed Architecture (IX)CERA-CRANIUM instantiation for UT2004
CERA sensory-motor services layer.CERA physical layer specialized processors.CERA mission-specific layer specialized p y pprocessors.CERA core layer goals and model stateCERA core layer goals and model state.
www.Conscious-Robots.com
Proposed Architecture (X)Example of percepts for UT2004
SP(Being Damaged) + SP(S E )SP(See Enemy) + CP(Enemy Approaching) =
MP(E Att ki )MP(Enemy Attacking)
www.Conscious-Robots.com
Evaluation (I)Assessing believability is not straighforward.
Subjective processSubjective process.Depends on opponents behavior.Game score is not significant.
AVERAGEPLAYER AVERAGE SCORE
Rule-Based System Bot 19.2
Q-Learning Bot 5.2
Q-Learning and Expert Systems Hybrid Bot 11.3
CERA CRANIUM Bot 18 3
www.Conscious-Robots.com
CERA-CRANIUM Bot 18.3
ConclusionsThe application of consciousness-inspired cognitive architectures is promisingcognitive architectures is promising.
W d i l i l i hWe need to incorporate learning algorithms and long term (episodic and semantic)
t th d lmemory to the model.
Improve ToM (opponent/friend modeling).
www.Conscious-Robots.com
Thank you
www.Conscious-Robots.com 22