s 1 - introduction rm fb

7/29/2019 S 1 - Introduction RM FB

1/22

Autonomous Mobile Robots, Chapter 1

R. Siegwart, I. Nourbakhsh

Autonomous Mobile Robots

The three key questions in Mobile Robotics

Where am I ?

Where am I going ?

How do I get there ?

To answer these questions the robot has to

have a model of the environment (given or autonomously built)

perceive and analyze the environment

find its position within the environment

plan and execute the movement

This course will deal with Locomotion and Navigation (Perception,

Localization, Planning and motion generation)

1

?



Content of the Course

1. Introduction

2. Locomotion

3. Mobile Robot Kinematics

4. Perception

5. Mobile Robot Localization

6. Planning and Navigation

Other Aspects of Autonomous Mobile Systems

Applications

1


2/22



ProgramDate Time Topic Responsible

22.3.07 14 - 16 Introduction: problem statements, typical applications, video R. Philippsen

29.3.07 14 - 16 Locomotion with legs and wheels (2h) R. Siegwart

5.4.07 14 - 16 Mobile Robots Kinematics I: Kinematics model (2h) R. Siegwart

5.4.07 16 - 18 Exercise 1: Kinematics model and trajectory calculation of wheeled robots R. Siegwart

12.4.07 14 - 16 Mobile Robots Kinematics II: Kinematics model (1h)

Mobile Robots Kinematics III: Motion control (1h)

A. Krebs

19.4.07 14 - 16 Perception II: Sensing and Perception (2h) R. Siegwart

19.4.07 16 - 18 Exercise 2: Motion control of a differentially driven robot, implementation on Matlab

and Lego-Robot

F. Tache

K. Macek

26.4.07 14 - 16 Perception III: Sensing and Perception, Uncertainty Representation S. Gchter

3.5.07 14 - 16 Feature extraction (1h)

Localization I: Introduction, odometry (1h)

V. Nguyen

3.5.07 16 - 18 Exercise 3: Vision and/or laser; take picture, feature extraction; uncertainty

representation; belief representation

D. Scaramuzza

S. Gchter

10.5.07 14 - 16 Localization II: Map representation, introduction to probabilistic map- based

localization, Markov localization(0.5h)

R. Triebel

24.5.07 14 - 16 Localization III: , Markov localization(1.5h), Kalman filter localization (0.5) R. Siegwart

24.5.07 16 - 18 Exercise 4: Partical Filter Based Localization, implementation on Matlab and Lego-

Robot

L. Spinello

X. Perrin

31.5.07 14 - 16 Localization IV: Kalman filter localization (1), Other examples of localization systems,map building (1)

R. Siegwart

31.5.07 16 - 18 Exercise 5: Probabilistic pose estimation with Khepera based on topological map,

implementation on Matlab and Lego-Robot

V. Nguyen

S. Vasudevan

7.6.07 14 - 16 Architectures for Navigation I: Introduction, path planning (2) R. Siegwart

14.6.07 14 - 16 Architectures for Navigation II: Obstacle avoidance, techniques for decomposition (2) R. Siegwart

14.6.07 16 - 18 Exercise 6: Obstacle avoidance base on local grid, implementation on Matlab and

Lego-Robot

A. Krebs

K. Macek

21.6.07 14 - 16 Summery and Outlook R. Siegwart



From Manipulators to Mobile Robots

1


3/22



Raw data

Environment ModelLocal Map

"Position"Global Map

Actuator Commands

Sensing Acting

InformationExtraction

PathExecution

CognitionPath Planning

Knowledge,Data Base

MissionCommands

Path

Real WorldEnvironment

LocalizationMap Building

Mo

tionControl

Perception

General Control Scheme for Mobile Robot Systems

1



Applications of Mobile Robots

Indoor Outdoor

Structured Environments Unstructured Environments

transportationindustry & service

cleaning ..large buildings

customer supportmuseums, shops ..

surveillance

buildings

research,entertainment,

toyunderwater

space

forestagriculture

construction

air

demining

mining

sewage tubes

fire fightingmilitary

1


4/22



Automatic Guided Vehicles

Newest generation ofAutomatic GuidedVehicle of VOLVO usedto transport motorblocks from onassembly station to another. It is guided by anelectrical wire installed

in the floor but it is alsoable to leave the wire toavoid obstacles. Thereare over 4000 AGV onlyat VOLVOs plants.

1



Helpmate

HELPMATE is a mobile robot used in hospitalsfor transportation tasks. It has various on boardsensors for autonomous navigation in the

corridors. The main sensor for localization is acamera looking to the ceiling. It can detect thelamps on the ceiling as reference (landmark).http://www.ntplx.net/~helpmate/

1


5/22



BR700 Cleaning Robot

BR 700 cleaning robotdeveloped and sold byKrcher Inc., Germany.Its navigation system isbased on a verysophisticated sonarsystem and a gyro.http://www.kaercher.de

1



ROV Tiburon Underwater Robot

Picture of robot ROV Tiburon forunderwater archaeology(teleoperated)- used by MBARI for

deep-sea research, this UAV providesautonomous hovering capabilities forthe human operator.

1


6/22



The Pioneer

Picture of Pioneer, the teleoperated robot that issupposed to explore the Sarcophagus atChernobyl

1



The Khepera Robot

KHEPERA is a small mobile robot for research and education. It sizes only about 60mm in diameter. Additional modules with cameras, grippers and much more areavailable. More then 700 units have already been sold (end of 1998).http://diwww.epfl.ch/lami/robots/K-family/ K-Team.html

1


7/22



Forester Robot

Pulstech developedthe first industrial likewalking robot. It isdesigned moving woodout of the forest. Theleg coordination is

automated, butnavigation is still doneby the human operatoron the robot.http://www.plustech.fi/

1



Robots for Tube Inspection

HCHER robots for sewage tubeinspection and reparation. Thesesystems are still fully teleoperated.http://www.haechler.ch

EPFL / SEDIREP: Ventilationinspection robot

1


8/22



Sojourner, First Robot on Mars

The mobile robotSojourner was usedduring the Pathfindermission to explorethe mars in summer1997. It was nearlyfully teleoperatedfrom earth. However,some on board

sensors allowed forobstacle detection.http://ranier.oact.hq.nasa.gov/telerobotics_page/telerobotics.shtm

1



NOMAD, Carnegie Mellon / NASAhttp://img.arc.nasa.gov/Nomad/

1


9/22



The Honda Walking Robot http://www.honda.co.jp/tech/other/robot.html

1



General Control Scheme for Mobile Robot Systems

1

Raw data


"Position"

Global Map

Actuator Commands

Sensing Acting


PathExecution


Knowledge,Data Base

MissionCommands

Path



MotionControl

Perception


10/22



Control Architectures / Strategies

Control Loop

dynamically changing

no compact model available

many sources of uncertainty

Two Approaches

Classical AI

o complete modeling

o function based

o horizontal decomposition

New AI, AL

o sparse or no modeling

o behavior based

o vertical decompositiono bottom up

1


Perception Motion Control

Cognition


Localization

PathEnvironment ModelLocal Map



Two Approaches

Classical AI(model based navigation)

complete modeling

function based

horizontal

decomposition

New AI, AL(behavior based navigation)

sparse or no modeling

behavior based

vertical decomposition

bottom up

Possible Solution

Combine Approaches

1


11/22



Mixed Approach Depicted into the General Control Scheme

1

Perception Motion Control

CognitionLocalization


Perceptionto

Action

Obstacle

Avoidance

Position

Feedback

Path

Environment Model

Local Map

Local Map

PositionPosition

Local Map



Environment Representation

Continuos Metric -> x,y,

Discrete Metric -> metric grid

Discrete Topological -> topological grid

Environment Modeling

Raw sensor data, e.g. laser range data, grayscale imageso large volume of data, low distinctiveness

o makes use of all acquired information

Low level features, e.g. line other geometric featureso medium volume of data, average distinctiveness

o filters out the useful information, still ambiguities

High level features, e.g. doors, a car, the Eiffel towero low volume of data, high distinctiveness

o filters out the useful information, few/no ambiguities, not enough information

Environment Representation and Modeling:

The Key for Autonomous Navigation

1


12/22



Odometry

not applicable

Modified

Environments

expensive,

inflexible

Feature-based

Navigation

still a challenge for

artificial systems

Environment Representation and Modeling: How we do it!

12195

34

39

25

Corridorcrossing

Elevator door

Entrance

Eiffel Tower

Landing at nightHow to find a treasure

1

CourtesyK.A

rras



Environment Representation: The Map Categories

Recognizable Locations Topological Maps

2 km

100 km

200 m

50 km

y

x{W}

Metric Topological Maps Fully Metric Maps (continuos ordiscrete)

1

Courtesy

K.

Arras


13/22



Environment Models: Continuous Discrete ; Raw data Features

Continuos

position in x,y,

Discrete

metric grid

topological grid

Raw Data

as perceived by sensor

A feature (or natural landmark) is anenvironmental structure which is static, alwaysperceptible with the current sensor system andlocally unique.

Examples

geometric elements (lines, walls, column ..)

a railway station

a river

the Eiffel Tower

a human being

fixed stars

skyscraper

1



Human Navigation: Topological with imprecise metric information

~ 400 m

~ 1 km

~ 200 m

~ 50 m

~ 10 m

1

Courtesy

K.

Arras


14/22



Incrementally

(dead reckoning)

Odometric or initialsensors (gyro)

not applicable

Modifying the environments

(artificial landmarks / beacons)

Inductive or optical tracks (AGV)

Reflectors or bar codes

expensive, inflexible

Methods for Navigation: Approaches with Limitations

1

CourtesyK.

Arras



Methods for Localization: The Quantitative Metric Approach

1. A priori Map: Graph, metric

2. Feature Extraction (e.g. line segments)

x

y

wxr

wyr

{W}

lwr

3. Matching:

Find correspondence

of features

4. Position Estimation:

e.g. Kalman filter, Markov

representation of uncertainties

optimal weighting acc. to a priori statistics

Odometry Observation

1

Courtesy

K.

Arras


15/22



Gaining Information through motion: (Multi-hypotheses tracking)

Believe state

1

Courtesy S. Thrun, W. Burgard



Grid-Based Metric Approach

Grid Map of the Smithsonians National Museum of American History in Washington DC.(Courtesy of Wolfram Burger et al.)

Grid: ~ 400 x 320 = 128000 points

1

Courtesy S. Thrun, W. Burgard


16/22



Methods for Localization: The Quantitative Topological Approach

1. A priori Map: Graphlocally uniquepoints

edges

2. Method for determiningthe local uniqueness

e.g. striking changes on raw data levelor highly distinctive features

3. Library of driving behaviors

e.g. wall or midline following, blind step,enter door, application specific

behaviorsExample: Video-based navigation with

natural landmarks

Courtesy of [Lanser et al. 1996]

1



Map Building: How to Establish a Map

1. By Hand

2. Automatically: Map Building

The robot learns its environment

Motivation:

- by hand: hard and costly- dynamically changing environment

- different look due to different perception

3. Basic Requirements of a Map:

a way to incorporate newly sensed

information into the existing world

model

information and procedures forestimating the robots position

information to dopath planning and

othernavigation task(e.g. obstacle

avoidance)

Measure of Quality of a map

topological correctness

metrical correctness

But: Most environments are a mixture ofpredictable and unpredictable features hybrid approach

model-based vs. behaviour-based

123.5

predictability

1

CourtesyK.

Arras


17/22



Map Building: The Problems

1. Map Maintaining: Keeping track ofchanges in the environment

e.g. disappearing

cupboard

- e.g. measure of belief of eachenvironment feature

2. Representation andReduction of Uncertainty

position of robot -> position of wall

position of wall -> position of robot

probability densities for feature positions

additional exploration strategies

?

1

CourtesyK.

Arras



Map Building: Exploration and Graph Construction

1. Exploration

- provides correct topology- must recognize already visited location

- backtracking for unexplored openings

2. Graph Construction

Where to put the nodes?

Topology-based: at distinctive locations

Metric-based: where features disappear orget visible

exploreon stack

alreadyexamined

1

CourtesyK.Arras


18/22



Control of Mobile Robots

Most functions forsave navigation are

local not involvinglocalization norcognition

Localization andglobal path planning

slower updaterate, only whenneeded

This approach ispretty similarto whathuman beings do.

global

local

1

Raw data



Actuator Commands

Sensing Acting


PathExecution


Knowledge,Data Base

MissionCommands

Path



Moti

onControl

Perception



Tour-Guide Robot (Nourbakhsh, CMU)

1


19/22



Autonomous Indoor Navigation (Thrun, CMU)

1



Tour-Guide Robot (EPFL @ expo.02)

1


20/22



Autonomous Indoor Navigation (Pygmalion EPFL)

very robust on-the-fly localization

one of the first systems with probabilistic sensor fusion

47 steps,78 meter length, realistic office environment,

conducted 16 times > 1km overall distance

partially difficult surfaces (laser), partially few vertical edges (vision)

1



Autonomous Robot for Planetary Exploration (ASL EPFL)

1


21/22



Humanoid Robots (Sony)

1



Turning Real Reality into VirtualReality

Courtesy of Sebastian Thrun

1


22/22



Robots Invading Our Environment

s 1 - introduction rm fb

Documents