Scene Based Reasoning Cognitive Architecture Frank Bergmann, [email protected] Brian Fenton,...
-
Upload
christine-rodgers -
Category
Documents
-
view
213 -
download
1
Transcript of Scene Based Reasoning Cognitive Architecture Frank Bergmann, [email protected] Brian Fenton,...
Scene Based ReasoningCognitive Architecture
Frank Bergmann, [email protected] Fenton, [email protected]://tinycog.sourceforge.net/
Reasoning
Scenes
Planner
Pla
nR
ecog
nitio
n
Pla
n
Exe
cutio
n
Persistent Plans
3D
Rec
onst
ruct
ion
Persistent Plans
Attention Subsystem
Persistent Plans
Episodic Memory
(cc) Frank Bergmann, Brian Fenton, http://tinycog.sourceforge.net/ 2
Problems Addressed We propose an integrated architecture for
implementing a number of “self-models”. We provide a model for talking about
modalities (auxiliary verbs) without the use of higher-order or modal logics.
Approach Use Thomas Metzinger’s “Self-Model
Theory of Subjectivity” as a kind of requirement statement.
Extend/generalize existing knowledge representation and reasoning formalisms to accommodate self-models.
Implementation Status Plays “Towers of Hanoi” Working on a “Stone age prey/hunter” sandbox http://tinycog.sourceforge.net/
Problems Addressed & Approach
Reasoning
Scenes
Planner
Pla
nR
ecog
nitio
n
Pla
n E
xecu
tion
Persistent Plans
3D
Rec
onst
ruct
ion
Persistent Plans
Attention Subsystem
Persistent Plans
Episodic Memory
(cc) Frank Bergmann, Brian Fenton, http://tinycog.sourceforge.net/ 3
Physical Volume Self-Model “This is the room I’m in, this is my leg.“Identify the 3D volume of the robot representing SBR and model its capabilities
Capabilities Self-Model „I am good at this.“Performance statistics of task decompositions
Planning Self-Model „I usually hit the goal in 30% of all cases.“Performance statistics of plans performed
Intention Self-Model ”I currently try to do this.“Introspection into the goals currently pursued
Goal Self-Model “I would like to be to do this.“Introspection into active “Persistent Goals”
Social Self-Model “Other group members respect me.“Role of self in group activities.
Behavioral Self-Model “I usually react like this.“Episodic Memory recordings of past SBR actions
Emotional Self-Model “I am happy to hear the news.”Introspection into current emotions and historic model of emotions
Historical Self-Model “I used to do a lot of this.”Episodic Memory recorded SBR past actions.
Terminological Self-Model “I know that a penguin is not a bird.”Export the Description Logic TBox into a 2d scene diagram.
Flow of Mental Events Self-Model “I just had an idea how to solve plan X”Short-term memory of “mental events”
Different Levels of Self-Models
Physical
Planning
Social
Internal
(cc) Frank Bergmann, Brian Fenton, http://tinycog.sourceforge.net/ 4
Scene Based Reasoning “Core” Architecture
Reasoning
Scenes
Planner
Pla
nR
ecog
nitio
n
Pla
n
Exe
cutio
n
Persistent Plans
3D
Rec
onst
ruct
ion
Persistent Plans
Attention Subsystem
Persistent Plans
Episodic Memory
1
2
3
20+ subsystems Explained at http://tinycog.sourceforge.net
(cc) Frank Bergmann, Brian Fenton, http://tinycog.sourceforge.net/ 5
SBR „Core“
Reasoning
„Scenes“
Planner
„Scene“
„Plan“ We use “Scenes” as a unified representation of internal and world states and as “situations” for the planner.– 3D Scene Graph level – suitable for 3D
reconstruction and low-level planning. Reasoning using physics simulation.Reasoning using “diagram reasoning”.
– Semantic Network level – “1st Mind”, provides “concepts” and “roles” as an abstraction from object and their attributes.Reasoning using Description Logic reasoning
– 2D Graph level – suitable to represent 2D maps, meta-representations of internal data-structure.
– “Situation” level – Scenes act as world states and internal states to the planner, adding “dynamics”
Eat DinnerStart Goal
. . .
. . .
. . .
Get pizzafrom Fridge
CleanUpEat
ObtainFood
Table1 Human1
. . .
. . .
Type: TableColor: greenPosition: ...Size: 1 x 1...
Type: HumanGender: malePosition: ......
in-front-of
Task/Action
Scene
1
2
3
1
Scene Based Reasoning “Core” Architecture
(cc) Frank Bergmann, Brian Fenton, http://tinycog.sourceforge.net/ 6
SBR „Core“
Reasoning
„Scenes“
Planner
„Scene“
„Plan“ We use “Scenes” as a unified representation of internal and world states and as “situations” for the planner.
Reasoning is performed by several specialized subsystems. – We use Description Logic as a “First
Mind” convenient “knowledge assembler“ to classify objects and their attributes into concepts and roles, but not for “higher-level” reasoning.
– 2D graph reasoning is a special type of planning.
– “Two Minds”: Planning together with the attention, motivational and “persistent goal” subsystems forms a 2nd reasoning system in addition to DL
Eat DinnerStart Goal
. . .
. . .
. . .
Get pizzafrom Fridge
CleanUpEat
ObtainFood
Table1 Human1
. . .
. . .
Type: TableColor: greenPosition: ...Size: 1 x 1...
Type: HumanGender: malePosition: ......
in-front-of
Task/Action
Scene
1
3
1
Scene Based Reasoning “Core” Architecture
2
2
(cc) Frank Bergmann, Brian Fenton, http://tinycog.sourceforge.net/ 7
SBR „Core“
Reasoning
„Scenes“
Planner
„Scene“
„Plan“ We use “Scenes” as a unified
representation of internal and world states and as “situations” for the planner.
Reasoning is performed by several specialized subsystems.
The „Planner“ uses Scenes as situations and goals.
– “Task decompositions” (HTNs) provide a middle ground between STRIPS style planning and procedural execution and are easy to learn.
– Actions may have multiple deterministic outcomes – statistics about actions are collected in the episodic memory.
– Physics simulations cover a range of planning capabilities that are difficult to handle using FOL and derivates
– A “simulation subsystem” allows for what-if analysis of plans
Eat DinnerStart Goal
. . .
. . .
. . .
Get pizzafrom Fridge
CleanUpEat
ObtainFood
Table1 Human1
. . .
. . .
Type: TableColor: greenPosition: ...Size: 1 x 1...
Type: HumanGender: malePosition: ......
in-front-of
Task/Action
Scene
2
1
3
1
2
3
Scene Based Reasoning “Core” Architecture
(cc) Frank Bergmann, Brian Fenton, http://tinycog.sourceforge.net/ 8
Attention Subsystem controls sensor focus Persistent Plans for setting plan priorities Episodic Memory stores plans indexed by content Plan Recognition allows for 1-shot learning 20+ subsystems defined at http://tinycog.sourceforge.net
Reasoning
Scenes
Planner
Pla
nR
ecog
nitio
n
Pla
n
Exe
cutio
n
Persistent Plans
3D
Rec
onst
ruct
ion
Persistent Plans
Attention Subsystem
Persistent Goal Hierarchy
Episodic Memory
Architecture Summary and “Two Minds”
(cc) Frank Bergmann, Brian Fenton, http://tinycog.sourceforge.net/ 9
Physical Volume Self-Model “This is my leg.“Identify the 3D volume of the robot representing SBR and model its capabilities
Capabilities Self-Model „I am good at this.“Performance statistics of task decompositions
Planning Self-Model „I usually hit the goal in 30% of all cases.“Performance statistics of plans performed
Intention Self-Model ”I currently try to do this.“Introspection into the goals currently pursued
Goal Self-Model “I would like to be to do this.“Introspection into active “Persistent Goals”
Social Self-Model “Other group members respect me.“Role of self in group activities.
Behavioral Self-Model “I usually react like this.“Episodic Memory recordings of past SBR actions
Emotional Self-Model “I am happy to hear the news.”Introspection into current emotions and historic model of emotions
Historical Self-Model “I used to do a lot of this.”Episodic Memory recorded SBR past actions.
Terminological Self-Model “I know that a penguin is not a bird.”Export the Description Logic TBox into a 2d scene diagram.
Flow of “Thoughts” Self-Model “I was surprised by this event”Short-term memory (“time-line”) of “mental events”
Self-Models and Self-Referentiality
(cc) Frank Bergmann, Brian Fenton, http://tinycog.sourceforge.net/ 10
Examples of Scenes representing Self-Models: Plan & DL-TBox
Modalities examples:– „My goal is ...“– „I believe that...“
Explicit representation of data-structures
No need to use modal or higher order logic
Apply the same reasoning engine as to object-level data
DinnerStart Goal
. . .
. . .
. . .
Get pizzafrom Fridge
CleanUpEat
ObtainFood
. . .
. . .
Bird
Canary
Penguin
Animal
Fish
has skin eats breathes moves
!can fly can swim
can sing is yellow
has wings can fly has
feathers has fins can swim has gills
(cc) Frank Bergmann, Brian Fenton, http://tinycog.sourceforge.net/ 11
Physical Volume Self-Model “This is my leg.“Identify the 3D volume of the robot representing SBR and model its capabilities
Capabilities Self-Model „I am good at this.“Performance statistics of task decompositions
Planning Self-Model „I usually hit the goal in 30% of all cases.“Performance statistics of plans performed
Intention Self-Model ”I currently try to do this.“Introspection into the goals currently pursued
Goal Self-Model “I would like to be to do this.“Introspection into active “Persistent Goals”
Social Self-Model “Other group members respect me.“Role of self in group activities.
Behavioral Self-Model “I usually react like this.“Episodic Memory recordings of past SBR actions
Emotional Self-Model “I am happy to hear the news.”Introspection into current emotions and historic model of emotions
Historical Self-Model “I used to do a lot of this.”Episodic Memory recorded SBR past actions.
Terminological Self-Model “I know that a penguin is not a bird.”Export the Description Logic TBox into a 2d scene diagram.
Flow of “Thoughts” Self-Model “I was surprised by this event”Short-term memory (“time-line”) of “mental events”
Self-Models and Self-Referentiality
Physical
Planning
Social
Internal