20
Integration of Integration of Representation Into Representation Into Goal-Driven Goal-Driven Behavior-Based Behavior-Based Robots Robots By Dr. Maja J. Mataric` By Dr. Maja J. Mataric` Presented by Andy Klempau Presented by Andy Klempau

Integration of Representation Into Goal-Driven Behavior-Based Robots

Embed Size (px)

DESCRIPTION

Integration of Representation Into Goal-Driven Behavior-Based Robots. By Dr. Maja J. Mataric` Presented by Andy Klempau. Introduction. Hybrid system separates reactive low level actions from deliberative decision actions. Alternative to Hybrid approach is Behavior-Based system. - PowerPoint PPT Presentation

Citation preview

Page 1: Integration of Representation Into Goal-Driven Behavior-Based Robots

Integration of Integration of Representation Into Goal-Representation Into Goal-

Driven Behavior-Based Driven Behavior-Based RobotsRobots

By Dr. Maja J. Mataric`By Dr. Maja J. Mataric`

Presented by Andy KlempauPresented by Andy Klempau

Page 2: Integration of Representation Into Goal-Driven Behavior-Based Robots

IntroductionIntroduction

Hybrid system Hybrid system separatesseparates reactive low level reactive low level actions from deliberative decision actions.actions from deliberative decision actions.

Alternative to Hybrid approach is Behavior-Alternative to Hybrid approach is Behavior-Based system.Based system.

Behavior-Based system Behavior-Based system combinescombines a reactive a reactive subsumption foundation with a decision maker.subsumption foundation with a decision maker.

Toto is a Behavior-Based robot that explores Toto is a Behavior-Based robot that explores dynamic office environments, identifies dynamic office environments, identifies landmarks, maps the landmarks, and uses the landmarks, maps the landmarks, and uses the map for path planning.map for path planning.

Page 3: Integration of Representation Into Goal-Driven Behavior-Based Robots

OutlineOutline

Toto’s StructureToto’s Structure

Basic Navigation ( Exploring )Basic Navigation ( Exploring )

Landmark DetectionLandmark Detection

Mapping LandmarksMapping Landmarks

Path PlanningPath Planning

Page 4: Integration of Representation Into Goal-Driven Behavior-Based Robots

Toto’s StructureToto’s Structure

Cylindrical RobotCylindrical Robot

12 Sonar Sensors12 Sonar Sensors

4 Bit Compass (16 states)4 Bit Compass (16 states)

Page 5: Integration of Representation Into Goal-Driven Behavior-Based Robots

Basic NavigationBasic Navigation

Subsumption ArchitectureSubsumption Architecture

Highest priority is Stroll behaviorHighest priority is Stroll behavior

Lowest priority is Correct behaviorLowest priority is Correct behavior

Page 6: Integration of Representation Into Goal-Driven Behavior-Based Robots

Basic NavigationBasic NavigationStroll :Stroll :

if( min( sonars 1 2 3 4 ) <= danger-zone if( min( sonars 1 2 3 4 ) <= danger-zone )) if( not stopped )if( not stopped )

stopstopelseelse

move backwardmove backward elseelse

move forwardmove forward

Page 7: Integration of Representation Into Goal-Driven Behavior-Based Robots

Basic NavigationBasic Navigation

Avoid :Avoid :if( ( sonar 3 or 4 ) <= safe-zone )if( ( sonar 3 or 4 ) <= safe-zone )

turn leftturn leftelse if( ( sonar 1 or 2 ) else if( ( sonar 1 or 2 )

<= safe-zone )<= safe-zone ) turn rightturn right

Page 8: Integration of Representation Into Goal-Driven Behavior-Based Robots

Basic NavigationBasic Navigation

Align :Align :

if( ( sonar 7 or 8 ) < edging-distance ANDif( ( sonar 7 or 8 ) < edging-distance AND( sonar 5 or 6 ) > edging-( sonar 5 or 6 ) > edging-

distance )distance ) turn rightturn rightif( ( sonar 9 or 10 ) < if( ( sonar 9 or 10 ) <

edging-distance ANDedging-distance AND ( sonar 11 or ( sonar 11 or 0 ) > edging-distance )0 ) > edging-distance ) turn leftturn left

Page 9: Integration of Representation Into Goal-Driven Behavior-Based Robots

Basic NavigationBasic Navigation

Correct :Correct :if( sonar 11 < edging-distance ANDif( sonar 11 < edging-distance AND

sonar 0 > edging-distance )sonar 0 > edging-distance )turn leftturn left

if( sonar 6 < edging-distance AND if( sonar 6 < edging-distance AND sonar 5 > edging-sonar 5 > edging-

distance )distance ) turn rightturn right

Page 10: Integration of Representation Into Goal-Driven Behavior-Based Robots

Landmark DetectionLandmark Detection

4 types of landmarks:4 types of landmarks: Right wall (RW); consistent right wall and consistent Right wall (RW); consistent right wall and consistent

direction.direction. Left wall (LW); consistent left wall and consistent Left wall (LW); consistent left wall and consistent

direction.direction. Corridor (C); consistent left and right walls and Corridor (C); consistent left and right walls and

consistent direction.consistent direction. Irregular (I); inconsistent walls and inconsistent Irregular (I); inconsistent walls and inconsistent

direction.direction.

How does Toto identify landmarks?How does Toto identify landmarks?ConfidenceConfidenceCounter!Counter!

Page 11: Integration of Representation Into Goal-Driven Behavior-Based Robots

Landmark DetectionLandmark Detection

After a time interval, sonar and compass After a time interval, sonar and compass readings are taken.readings are taken.Confidence Counter increments when Confidence Counter increments when sonar and compass readings are the same sonar and compass readings are the same as last time interval.as last time interval.Predetermined threshold identifies how Predetermined threshold identifies how many time intervals are needed to justify a many time intervals are needed to justify a landmarklandmark

Page 12: Integration of Representation Into Goal-Driven Behavior-Based Robots

Landmark DetectionLandmark Detection

Page 13: Integration of Representation Into Goal-Driven Behavior-Based Robots

Mapping LandmarksMapping Landmarks

After discovered, landmarks are stored in After discovered, landmarks are stored in Toto’s internal map.Toto’s internal map.

Landmark nodes store information discovered Landmark nodes store information discovered through sensors and compass (through sensors and compass (see next slidesee next slide).).

Nodes communicate with neighbors.Nodes communicate with neighbors.

Page 14: Integration of Representation Into Goal-Driven Behavior-Based Robots

Mapping LandmarksMapping Landmarks

Landmark node has a set < T, C, L, P > whereLandmark node has a set < T, C, L, P > where

T is { LW, RW, C, I }; qualitative landmark type.T is { LW, RW, C, I }; qualitative landmark type.

C is [ 0 … 15 ]; averaged compass bearing.C is [ 0 … 15 ]; averaged compass bearing.

L is [ 1 … 127 ]; rough estimate of landmark’s length.L is [ 1 … 127 ]; rough estimate of landmark’s length.

P = ( x, y ) -128 <= x, y <= 127; coarse position estimate.P = ( x, y ) -128 <= x, y <= 127; coarse position estimate.

Length is obtained through timer Length is obtained through timer (could be confidence (could be confidence

counter)counter)..

Position is obtained through length and Position is obtained through length and compass.compass.

Page 15: Integration of Representation Into Goal-Driven Behavior-Based Robots

Mapping LandmarksMapping Landmarks

Example:Example:

Page 16: Integration of Representation Into Goal-Driven Behavior-Based Robots

Path PlanningPath Planning

Use the map to go to a goal.Use the map to go to a goal.

This is done by activating one of Toto’s This is done by activating one of Toto’s previously visited landmarks as a goal.previously visited landmarks as a goal.

Page 17: Integration of Representation Into Goal-Driven Behavior-Based Robots

Path PlanningPath Planning

Goal sends signal to neighbor nodes.Goal sends signal to neighbor nodes.

Eventually, all nodes know where goal is.Eventually, all nodes know where goal is.

Greedy algorithm ensures Toto will take shortest path to goal.Greedy algorithm ensures Toto will take shortest path to goal.

Toto can go to goal starting from any landmark.Toto can go to goal starting from any landmark.

Toto can adapt to a changing environment.Toto can adapt to a changing environment.

Page 18: Integration of Representation Into Goal-Driven Behavior-Based Robots

ReviewReview

ExploresExplores

Finds landmarksFinds landmarks

Stores landmarks in Stores landmarks in mapmap

Goes to goalGoes to goal

Page 19: Integration of Representation Into Goal-Driven Behavior-Based Robots

ConclusionConclusion

Toto explores, maps, plans, and finds Toto explores, maps, plans, and finds goals goals withoutwithout Deliberative or Hybrid Deliberative or Hybrid process.process.

Toto “extends the repertoire of integrated Toto “extends the repertoire of integrated reactive systems to tasks requiring spatial reactive systems to tasks requiring spatial modeling and user interaction.”modeling and user interaction.”

Toto can adapt to a dynamic environment.Toto can adapt to a dynamic environment.

Page 20: Integration of Representation Into Goal-Driven Behavior-Based Robots

DiscussionDiscussion

Is linear-time path planning reactive?Is linear-time path planning reactive?

Can a Behavior-Based system do anything Can a Behavior-Based system do anything a hybrid system can do?a hybrid system can do?

How is the open-space behavior How is the open-space behavior triggered? triggered?