Robot Mapping

Introduction to Robot Mapping

Gian Diego Tipaldi, Wolfram Burgard

1
What is Robot Mapping?
 Robot – a device, that moves through
the environment

 Mapping – modeling the environment

2
Related Terms

State
Localization
Estimation

Mapping SLAM

Motion
Navigation
Planning
3
What is SLAM?
 Computing the robot’s poses and the
map of the environment at the same
time

 Localization: estimating the robot’s

location
 Mapping: building a map
 SLAM: building a map and localizing
the robot simultaneously
4
Localization Example
 Estimate the robot’s poses given
landmarks

Courtesy: M. Montemerlo 5
Mapping Example
 Estimate the landmarks given the
robot’s poses

Courtesy: M. Montemerlo 6
SLAM Example
 Estimate the robot’s poses and the
landmarks at the same time

Courtesy: M. Montemerlo 7
The SLAM Problem
 SLAM is a chicken-or-egg problem:
→ a map is needed for localization and
→ a pose estimate is needed for mapping

map

localize
8
SLAM is Relevant
 It is considered a fundamental
problem for truly autonomous robots
 SLAM is the basis for most navigation
systems

map

autonomous
navigation

localize
9
SLAM Applications
 SLAM is central to a range of indoor,
outdoor, air and underwater applications
for both manned and autonomous vehicles.

Examples:
 At home: vacuum cleaner, lawn mower
 Air: surveillance with unmanned air vehicles
 Underwater: reef monitoring
 Underground: exploration of mines
 Space: terrain mapping for localization
10
SLAM Applications
Indoors Undersea

Space Underground

Courtesy: Evolution Robotics, H. Durrant-Whyte, NASA, S. Thrun 11

SLAM Showcase – Mint

Courtesy: Evolution Robotics (now iRobot) 12

Mapping Freiburg CS Campus

14
Definition of the SLAM Problem
Given
 The robot’s controls

 Observations

Wanted
 Map of the environment

 Path of the robot

15
Probabilistic Approaches
 Uncertainty in the robot’s motions and
observations
 Use the probability theory to explicitly
represent the uncertainty

“The robot is “The robot is

exactly here” somewhere here” 16
In the Probabilistic World
Estimate the robot’s path and the map

distribution path map given observations controls

17
Graphical Model

unknown

observed

unknown

Courtesy: Thrun, Burgard, Fox 18

Full SLAM vs. Online SLAM
 Full SLAM estimates the entire path

 Online SLAM seeks to recover only the

most recent pose

19
Graphical Model of Online SLAM

Courtesy: Thrun, Burgard, Fox 20

Online SLAM
 Online SLAM means marginalizing out
the previous poses

 Integrals are typically solved

recursively, one at at time

21
Graphical Model of Online SLAM

Courtesy: Thrun, Burgard, Fox 22

Why is SLAM a Hard Problem?
1. Robot path and map are both unknown

2. Map and pose estimates correlated

Courtesy: M. Montemerlo 23
Why is SLAM a Hard Problem?
 The mapping between observations and
the map is unknown
 Picking wrong data associations can have
catastrophic consequences (divergence)

Robot pose
uncertainty

Courtesy: M. Montemerlo 24
Taxonomy of the SLAM Problem
Volumetric vs. feature-based SLAM

Courtesy: D. Hähnel Courtesy: E. Nebot 25

Taxonomy of the SLAM Problem
Topologic vs. geometric maps

26
Taxonomy of the SLAM Problem
Known vs. unknown correspondence

27
Taxonomy of the SLAM Problem
Static vs. dynamic environments

28
Taxonomy of the SLAM Problem
Small vs. large uncertainty

29
Taxonomy of the SLAM Problem
Active vs. passive SLAM

Image courtesy by Petter Duvander

30
Taxonomy of the SLAM Problem
Any-time and any-space SLAM

31
Taxonomy of the SLAM Problem
Single-robot vs. multi-robot SLAM

32
Approaches to SLAM
 Large variety of different SLAM
approaches have been proposed
 Most robotics conferences dedicate
multiple tracks to SLAM
 The majority of techniques uses
probabilistic concepts
 History of SLAM dates back to the
mid-eighties
 Related problems in geodesy and
photogrammetry
33
SLAM History by Durrant-Whyte
 1985/86: Smith et al. and Durrant-Whyte
describe geometric uncertainty and
relationships between features or landmarks
 1986: Discussions at ICRA on how to solve
the SLAM problem followed by the key
paper by Smith, Self and Cheeseman
 1990-95: Kalman-filter based approaches
 1995: SLAM acronym coined at ISRR’95
 1995-1999: Convergence proofs & first
demonstrations of real systems
 2000: Wide interest in SLAM started
34
Three Main Paradigms

Kalman Particle Graph-

filter filter based

35
Motion and Observation Model

"Motion model"

"Observation model"

Courtesy: Thrun, Burgard, Fox 36

Motion Model
 The motion model describes the
relative motion of the robot

distribution new pose given old pose control

37
Motion Model Examples
 Gaussian model

 Non-Gaussian model

Courtesy: Thrun, Burgard, Fox 38

More on Motion Models
 Course: Introduction to Mobile
Robotics, Chapter 6
 Thrun et al. “Probabilistic Robotics”,
Chapter 5

40
Observation Model
 The observation or sensor model
relates measurements with the robot’s
pose

distribution observation given pose

41
Observation Model Examples
 Gaussian model

 Non-Gaussian model

42
More on Observation Models
 Course: Introduction to Mobile
Robotics, Chapter 7
 Thrun et al. “Probabilistic Robotics”,
Chapter 6

43
Summary
 Mapping is the task of modeling the
environment
 Localization means estimating the
robot’s pose
 SLAM = simultaneous localization and
mapping
 Full SLAM vs. Online SLAM
 Rich taxonomy of the SLAM problem

44
Literature
SLAM overview
 Springer “Handbook on Robotics”,
Chapter on Simultaneous Localization
and Mapping (subsection 1 & 2)
On motion and observation models
 Thrun et al. “Probabilistic Robotics”,
Chapters 5 & 6
 Course: Introduction to Mobile
Robotics, Chapters 6 & 7
45
46
Slide Information
 These slides have been created by Cyrill Stachniss as part of
the robot mapping course taught in 2012/13 and 2013/14.
 I tried to acknowledge all people that contributed image or
video material. In case I missed something, please let me
know. If you adapt this course material, please make sure
you keep the acknowledgements.
 Feel free to use and change the slides. If you use them, I
would appreciate an acknowledgement as well. To satisfy my
own curiosity, I appreciate a short email notice in case you
use the material in your course.
 My video recordings are available through YouTube:
http://www.youtube.com/playlist?list=PLgnQpQtFTOGQrZ4O5QzbIHgl3b1JHimN_&feature=g-list

Cyrill Stachniss, 2014

[email protected]
bonn.de47