probabilistic algorithms for mobile robot mapping
Download
Skip this Video
Download Presentation
Probabilistic Algorithms for Mobile Robot Mapping

Loading in 2 Seconds...

play fullscreen
1 / 52

Probabilistic Algorithms for Mobile Robot Mapping - PowerPoint PPT Presentation


  • 371 Views
  • Uploaded on

Probabilistic Algorithms for Mobile Robot Mapping. Sebastian Thrun Carnegie Mellon & Stanford Wolfram Burgard University of Freiburg and Dieter Fox University of Washington. Based on the paper A Real-Time Algorithm for Mobile Robot Mapping With Applications to Multi-Robot and 3D Mapping

loader
I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
capcha
Download Presentation

PowerPoint Slideshow about 'Probabilistic Algorithms for Mobile Robot Mapping' - Rita


An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
probabilistic algorithms for mobile robot mapping

Probabilistic Algorithms forMobile Robot Mapping

Sebastian Thrun

Carnegie Mellon & Stanford

Wolfram Burgard

University of Freiburg

and Dieter Fox

University of Washington

slide2
Based on the paper

A Real-Time Algorithm for Mobile Robot Mapping

With Applications to Multi-Robot and 3D Mapping

Best paper award at 2000 IEEE International Conference on Robotics

and Automation (~1,100 submissions)

Sponsored by DARPA (TMR-J.Blitch, MARS-D.Gage, MICA-S.Heise)

and NSF (ITR(2), CAREER-E.Glinert, IIS-V.Lumelsky)

Other contributors: Yufeng Liu, Rosemary Emery, Deepayan Charkrabarti, Frank Dellaert, Michael

Montemerlo, Reid Simmons, Hugh Durrant-Whyte, Somajyoti Majnuder, Nick Roy, Joelle Pineau, …

slide3

This Talk

Motivation

SLAM

(Kalman filters)

Expectation

Maximization

Real Time

Hybrid

3D Mapping

with EM

Open

Problems

museum tour guide robots
Museum Tour-Guide Robots

With: Greg Armstrong, Michael Beetz, Maren Benewitz, Wolfram Burgard, Armin Cremers, Frank Dellaert, Dieter Fox, Dirk Haenel, Chuck Rosenberg, Nicholas Roy, Jamie Schulte, Dirk Schulz

the nursebot initiative
The Nursebot Initiative

With: Greg Armstrong, Greg Baltus, Jacqueline Dunbar-Jacob, Jennifer Goetz, Sara Kiesler, Judith Matthews, Colleen McCarthy, Michael Montemerlo, Joelle Pineau, Martha Pollack, Nicholas Roy, Jamie Schulte

mapping the problem
Mapping: The Problem
  • Concurrent Mapping and Localization (CML)
  • Simultaneous Localization and Mapping (SLAM)
mapping the problem1
Mapping: The Problem
  • Continuous variables
  • High-dimensional (eg, 1,000,000+ dimensions)
  • Multiple sources of noise
  • Simulation not acceptable
milestone approaches
Milestone Approaches

Mataric 1990

Elfes/Moravec 1986

Kuipers et al 1991

Lu/Milios/Gutmann 1997

3d mapping
3D Mapping

Moravec et al, 2000

Konolige et al, 2001

Teller et al, 2000

take home message

Every state-of-the-art

mapping algorithm

is probabilistic.

Take-Home Message

Mapping is the

holy grail in

mobile robotics.

slide12

This Talk

Motivation

SLAM

(Kalman filters)

Expectation

Maximization

Real Time

Hybrid

3D Mapping

with EM

Open

Problems

bayes filters
Bayes Filters

x = state

t = time

z = measurement

u = control

 = constant

Special cases: HMMs

DBNs

POMDPs

Kalman filters

Condensation

...

bayes filters in localization
Bayes Filters in Localization

[Simmons/Koenig 95]

[Kaelbling et al 96]

[Burgard, Fox, et al 96]

bayes filters for mapping

s = robot pose

m = map

t = time

 = constant

z = measurement

u = control

Localization:

Mapping?

Bayes Filters for Mapping
kalman filters slam
Kalman Filters (SLAM)

[Smith, Self, Cheeseman, 1990]

slam limitations
SLAM: Limitations
  • Linear
  • Scaling: O(N4) in number of features in map
  • Can’t solve data association problem
slide20

This Talk

Motivation

SLAM

(Kalman filters)

Expectation

Maximization

Real Time

Hybrid

3D Mapping

with EM

Open

Problems

unknown data association em

E-Step: Localization

M-Step: Mapping with known poses

Unknown Data Association: EM

[Dempster et al, 77] [Thrun et al, 1998] [Shatkay/Kaelbling 1997]

cmu s wean hall 80 x 25 meters

16 landmarks

15 landmarks

27 landmarks

17 landmarks

CMU’s Wean Hall (80 x 25 meters)
em mapping limitations
EM Mapping: Limitations
  • Local Minima
  • Not Real-Time
slide25

This Talk

Motivation

SLAM

(Kalman filters)

Expectation

Maximization

Real Time

Hybrid

3D Mapping

with EM

Open

Problems

the goal
The Goal

Kalman filters:

real-time

No data association

EM:

data association

Not real-time

?

real time approximation

+

Real-Time Approximation

Our ICRA Paper 

real time approximation1
Real-Time Approximation

Yellow flashes:

artificially distorted map (30 deg, 50 cm)

importance of posterior pose estimate
Importance of Posterior Pose Estimate

With pose posterior

Without pose posterior

multi robot mapping
Multi-Robot Mapping

Cascaded architecture

  • Every module maximizes likelihood
  • Pre-aligned scans are passed up in hierarchy

map

map

Pre-aligned scans

Aligned map

map

map

map

multi robot exploration
Multi-Robot Exploration

DARPA TMR Texas 7/99

(July. Texas. No air conditioning.

Req to dress up. Rattlesnakes)

DARPA TMR Maryland 7/00

slide40

This Talk

Motivation

SLAM

(Kalman filters)

Expectation

Maximization

Real Time

Hybrid

3D Mapping

with EM

Open

Problems

multi planar 3d mapping
Multi-Planar 3D Mapping

Idea: Exploit fact that buildings posses many planar surfaces

  • Compact models
  • High Accuracy
  • Objects instead of pixels
3d multi plane mapping problem
3D Multi-Plane Mapping Problem

Entails five problems

  • Generative model with priors: Not all of the world is planar
  • Parameter estimation: Location and angle of planar surfaces unknown
  • Outlier identification: Not all measurements correspond to planar surfaces (other objects, noise)
  • Correspondence: Different measurements correspond to different planar surfaces
  • Model selection: Number of planar surfaces unknown
em to the rescue

*

*

*

*

*

*

EM To The Rescue!

Game Over!

results
Results

With EM

(95% of data explained by 7 surfaces)

Without EM

error

With: Deepayan Chakrabarti, Rosemary Emery, Yufeng Liu, Wolfram Burgard, ICML-01

the obvious next step
The Obvious Next Step

EM for

concurrent

localization

EM for

object

mapping

underwater mapping with university of sydney
Underwater Mapping (with University of Sydney)

With: Hugh Durrant-Whyte, Somajyoti Majunder, Marc de Battista, Steve Scheding

slide48

This Talk

Motivation

SLAM

(Kalman filters)

Expectation

Maximization

Real Time

Hybrid

3D Mapping

with EM

Open

Problems

take home message1
Take-Home Message

Mapping is the

holy grail in

mobile robotics.

Every state-of-the-art

mapping algorithm

is probabilistic.

Sebastian has

one cool animation!

open problems
Open Problems
  • 2D Indoor mapping and exploration
  • 3D mapping (real-time, multi-robot)
  • Object mapping (desks, chairs, doors, …)
  • Outdoors, underwater, planetary
  • Dynamic environments (people, retail stores)
  • Full posterior with data association (real-time, optimal)
open problems con t
Open Problems, con’t
  • Mapping, localization
  • Control/Planning under uncertainty
  • Integration of symbolic making
  • Human robot interaction

Literature Pointers:

  • “Robotic Mapping” at www.thrun.org
  • “Probabilistic Robotics” AI Magazine 21(4)
ad