Thông tin tài liệu
Multi-Robot S
y
stems. From Swarms to Intelli
g
ent Automata
V
olume III
Volume III
Proceedings from the 2005 International Workshop
on Multi-Robot Systems
Edited b
y
LYNNE E. PARKER
T
h
e University o
f
Tennessee
,
Knoxvi
ll
e, TN, U.S.A
.
an
d
FRANK E. SCHNEIDER
Multi-Robot Systems.
From
Swarms to
Intelligent
Automata
ALAN C. SCHULTZ
N
av
y
Center
f
or Applied Research in A.I.
,
N
ava
l
Researc
h
La
b
oratory
,
Was
h
ington, DC, U.S.A
.
F
GAN, Wac
h
t
b
erg, Germany
A
C.I.P. Catalogue record for this book is available from the Library of Congress.
Published by Springer,
P.O. Box 17
,
3300 AA Dordrecht
,
The Netherlands.
Printed on acid-
f
ree pape
r
All Rights Reserved
©
2005 Sprin
g
e
r
No part of this work may be reproduced, stored in a retrieval system, or transmitted
i
n any form or by any means, electronic, mechanical, photocopying, microfilming,
r
ecording or otherwise, without written permission from the Publisher, with the
exception of an
y
material supplied specificall
y
for the purpose of bein
g
entered
and executed on a computer system, for exclusive use by the purchaser of the work.
Print
ed
in th
e
N
e
th
e
rlan
ds.
ISBN-13 978-1-4020-3388-9 (HB) Springer Dordrecht, Berlin, Heidelberg, New York
ISBN-10 1-4020-3388-5 (HB) Sprin
g
er Dordrecht, Berlin, Heidelber
g
, New York
ISBN-10 1-4020-3389-3 (e-book) Springer Dordrecht, Berlin, Heidelberg, New York
I
SBN-13 978-1-4020-3389-6 (e-book) Springer Dordrecht, Berlin, Heidelberg, New York
C
ontents
Pr
e
f
ace
ix
Part I Task Allocatio
n
The Generation of Bidding Rules for Auction-Based Robot Coordination
3
C
rai
g
Tove
y
, Michail G. La
g
oudaki
s
,
Sonal Jain, and Sven Koeni
g
Issues in Multi-Robot Coalition Formatio
n
15
L
ovekesh Vi
g
and Julie A. Adam
s
Sensor Network-Mediated Multi-Robot Task Allocatio
n
27
M
axim A. Batalin and
G
aura
vS
.
S
ukhatm
e
Part II Coordination in Dynamic Environment
s
M
ulti-Ob
j
ective Cooperative Control of D
y
namical S
y
stem
s
41
Z
hihua Q
u
,
Jin
g
Wan
g
,
a
nd Richard A. Hul
l
L
evels of Multi-Robot Coordination for D
y
namic Environment
s
53
C
olin P. McMillen, Paul E. R
y
bski, and Manuela M. Velos
o
Parallel Stochastic Hill-Climbin
g
with Small Team
s
6
5
B
rian P. Gerke
y
, Sebastian Thru
n
,
Geo
ff
Gordo
n
T
owar
d
Versat
ili
t
y
o
f
Mu
l
t
i
-Ro
b
ot S
y
stem
s
79
C
o
l
in C
h
err
y
an
d
Hon
g
Z
h
an
g
Part III In
f
ormat
i
on / Sensor S
h
ar
i
n
g
an
d
Fus
i
o
n
Decentra
li
ze
d
Commun
i
cat
i
on Strate
gi
es
f
or Coor
di
nate
d
Mu
l
t
i
-A
g
ent Po
li
c
i
es 9
3
M
aayan Rot
h
, Rei
d
Simmons, an
d
Manue
l
aVe
l
os
o
Improving Multirobot Multitarget Tracking by Communicating
Ne
g
at
i
ve In
f
ormat
i
o
n
pgg
107
M
att
h
ew Powers, Ramprasa
d
Ravic
h
an
d
ran, Fran
k
De
ll
aert, an
d
Tuc
k
er Ba
l
c
h
vi
MU
LTI-R
O
B
O
T
S
Y
S
TEM
S
Enabling Autonomous Sensor-Sharing for Tightly-Coupled
C
ooperat
i
ve Tas
k
s
g
119
Ly
nne E. Parker, Maureen Chandra, and Fan
g
Tan
g
Part IV Distributed Mapping and Coverag
e
Merging Partial Maps without Using Odometr
y
1
3
3
Distributed Coverage of Unknown/Unstructured Environments
b
yMo
bil
e Sensor Networ
ks
g
14
5
P
art V Motion Planning and Contro
l
1
59
J
ames Bruce and Manuela
V
elos
o
A Multi-Robot Testbed for Biologically-Inspired
C
ooperat
i
ve Contro
l
171
Rafael Fierro, Justin Clark
,
k
k
Dean Hou
g
en, and Sesh Commuri
P
art VI Human-Robot Interactio
n
T
ask Switching and Multi-Robot Team
s
1
8
5
Michael A.
G
oodric
h
,
Mor
g
an Qui
g
le
y
,
a
nd Ker
y
l Cosenz
o
User Modelling for Principled Sliding Autonomy in Human-Robot Teams 19
7
Brennan Sellner, Reid Simmons, and San
j
iv Sing
h
P
art VII A
pp
lication
s
Multi-Robot Chemical Plume Tracin
g
211
Diana Spears, Dimitri Zarzhitsk
y
,
a
nd David Tha
y
er
Deploying Air-Ground Multi-Robot Teams
i
nUr
b
an Env
i
ronment
s
pyg
pyg
223
L. Chaimowicz, A. Cowle
y
, D. Gomez-Ibanez, B. Grocholsk
y
, M. A. Hsieh
,
H. Hsu, J
.
F. Keller, V. Kumar, R. Swaminathan, and C. J. Ta
y
lo
r
P
art VIII Poster S
h
ort Paper
s
A
Robust Monte-Carlo Algorithm for Multi-Robot Localizatio
n
251
A
Dialogue-Based Approach to Multi-Robot Team Contro
l
2
5
7
N
athanael Chambers, James Allen, Lucian Galescu, and Hyuckchul Jun
g
F
r
F
F
ancesco
rr
A
mi
g
oni
,
S
imon
e
G
as
p
arini, and Maria Gin
i
I
oanni
s
Rekleiti
s
,
Ai
P
eng
PP
N
ew
,
and Howie Choset
R
eal-Time Multi-Robot Motion Planning with Safe Dynamic
s
Va
z
ha Amiranashvil
i
and Gerhard Lakeme
y
e
r
Prec
i
s
i
on Man
ip
u
l
at
i
on w
i
t
h
Coo
p
erat
i
ve Ro
b
ot
s
2
35
A
s
h
l
e
y
e
S
t
r
o
r
r
u
p
u
e
,
T
e
T
T
r
r
y
r
r
H
u
HH
n
t
s
tt
b
e
r
g
rr
e
r
,
r
r
A
v
i
O
k
o
n
, and Hrand Aghazarian
C
ontent
s
vii
for Mobile Robot Teams
263
J
ason Derenick, Christo
p
her Thorne, and John S
p
letze
r
T
h
e
G
NATs – Lo
w
-
C
ost Em
b
e
dd
e
d
Net
w
or
ks
f
or Support
i
n
g
Mo
bil
eRo
b
ot
s
277
Keit
h
J. O’Hara, Danie
l
B. Wa
lk
er, an
d
Tuc
k
er R. Ba
l
c
h
2
9
1
2
9
9
S
warm
i
n
g
UAVS Be
h
av
i
or H
i
erarc
hy
269
K
uo-
C
hi Li
n
Ro
l
eBase
d
Operat
i
on
s
283
B
rian Satterˇ eld
,
Heeten Choxi
,
and Drew Housten
Hybrid
f
Free-Space Optics/Radio Frequency (FSO/RF) Networks
bil b
b
Er
godic Dynamics by Design: A Route to Predictable Multi-Robot System
s
A
ut
h
or
In
de
x
Dy
la
n
A.
S
hell
,
C
hri
s
V
V
V
J
ones, and Maja J. Matari
JJ
c
´
Prefac
e
T
h
eT
hi
r
d
Internat
i
ona
l
Wor
k
s
h
op on Mu
l
t
i
-Ro
b
ot Systems was
h
e
ld in
March 200
5
at the Naval Research Laboratory in Washington, D.C., USA
.
Br
i
ng
i
ng toget
h
er
l
ea
di
ng researc
h
ers an
d
government sponsors
f
or t
h
ree
d
ay
s
of
tec
h
n
i
ca
li
nterc
h
ange on mu
l
t
i
-ro
b
ot systems, t
h
ewor
k
s
h
op
f
o
ll
ows tw
o
p
rev
i
ous
hi
g
hl
y success
f
u
l
gat
h
er
i
ngs
i
n 2002 an
d
2003.L
ik
et
h
e prev
i
ous tw
o
wor
k
s
h
ops, t
h
e meet
i
ng
b
egan w
i
t
h
presentat
i
ons
b
yvar
i
ous government pro
-
gram managers
d
escr
ibi
ng app
li
cat
i
on areas an
d
programs w
i
t
h
an
i
nterest
in
m
u
l
t
i
-ro
b
ot systems. U.S. Government representat
i
ves were on
h
an
df
rom
t
h
eO
ffi
ce o
f
Nava
l
Researc
h
an
d
severa
l
ot
h
er governmenta
l
o
ffi
ces.Top re
-
searc
h
ers
i
nt
h
e
fi
e
ld
t
h
en presente
d
t
h
e
i
r current act
i
v
i
t
i
es
i
n many areas o
f
m
u
l
t
i
-ro
b
ot s
y
stems. Presentat
i
ons spanne
d
aw
id
e ran
g
eo
f
top
i
cs,
i
nc
l
u
d
-
i
n
g
tas
k
a
ll
ocat
i
on, coor
di
nat
i
on
i
n
dy
nam
i
cenv
i
ronments,
i
n
f
ormat
i
on/senso
r
s
h
ar
i
n
g
an
df
us
i
on,
di
str
ib
ute
d
mapp
i
n
g
an
d
covera
g
e, mot
i
on p
l
ann
i
n
g
an
d
c
ontro
l
,
h
uman-ro
b
ot
i
nteract
i
on, an
d
app
li
cat
i
ons o
f
mu
l
t
i
-ro
b
ot s
y
stems. A
ll
p
resentations were
g
iven in a sin
g
le-track workshop format. This proceed
-
i
n
g
s documents the work presented at the workshop.The research presenta
-
tions were followed b
y
panel discussions, in which all participants interacte
d
to hi
g
hli
g
ht the challen
g
es of this field and to develop possible solutions. I
n
addition to the invited research talks, researchers and students were
g
iven a
n
o
pportunit
y
to present their work at poster sessions.We would like to thank th
e
Naval Research Laborator
y
for sponsorin
g
this workshop and providin
g
the fa-
c
ilities for these meetin
g
s to take place.We are extremel
yg
rateful to Ma
g
dalen
a
Bu
g
a
j
ska, Paul Wie
g
and, and Mitchell A. Potter, for their vital help (and lon
g
hours) in editin
g
these proceedin
g
s and to Michelle Caccivio for providin
g
th
e
administrative su
pp
ort to the worksho
p
.
L
YNNE
E
.
P
ARKER
,
A
L
AN
C
.
S
C
H
U
LT
Z
,
A
ND
F
R
A
N
K
E
.
S
C
HNEIDER
ix
I
T
A
S
K ALL
OC
ATI
ON
THE GENERATION OF BIDDING RULES FOR
AUCTION-BASED ROBOT COORDINATION
∗
C
ra
i
g Tovey, M
i
c
h
a
il
G. Lagou
d
a
kis
Sc
h
oo
l
of In
d
ustria
l
an
d
Systems Engineering, Georgia Institute of Tec
h
no
l
og
y
{
ctovey, m
i
c
h
a
il
.
l
a
g
ou
d
a
kis
}
@
isye.
g
atech.ed
u
S
ona
l
Ja
i
n, Sven Koen
i
g
Computer Science Department, University of Sout
h
ern Ca
l
iforni
a
{
s
ona
lj
a
i
,s
k
oen
ig
}
@
usc.ed
u
Abs
tr
act
R
o
b
ot
i
cs researc
h
ers
h
ave use
d
auct
i
on-
b
ase
d
coor
di
nat
i
on systems
f
or ro
b
o
t
t
eams because of their robustness and efficiency. However, there is no researc
h
i
nto systematic methods for deriving appropriate bidding rules for given tea
m
o
bjectives. In this paper, we propose the first such method and demonstrate it b
y
d
eriving bidding rules for three possible team objectives of a multi-robot explo
-
r
ation task. We demonstrate experimentally that the resulting bidding rules in
-
d
eed exhibit good performance for their respective team objectives and compar
e
f
avorably to the optimal performance. Our research thus allows the designer
s
o
f auction-based coordination systems to focus on developing appropriate tea
m
o
bjectives, for which good bidding rules can then be derived automatically
.
K
eywords:
A
uctions, Bidding Rules, Multi-Robot Coordination, Exploration
.
1. Introduction
T
h
et
i
me requ
i
re
d
to reac
h
ot
h
er p
l
anets ma
k
es p
l
anetary sur
f
ace exp
l
orat
i
o
n
mi
ss
i
ons pr
i
me targets
f
or automat
i
on. Sen
di
ng rovers to ot
h
er p
l
anets e
i
t
h
e
r
i
nstea
d
o
f
or toget
h
er w
i
t
h
peop
l
e can a
l
so s
i
gn
ifi
cant
l
yre
d
uce t
h
e
d
anger an
d
c
ost
i
nvo
l
ve
d
. Teams o
f
rovers are
b
ot
h
more
f
au
l
tto
l
erant (t
h
roug
h
re
d
un
-
d
ancy) an
d
more e
ffi
c
i
ent (t
h
roug
h
para
ll
e
li
sm) t
h
an s
i
ng
l
e rovers
if
t
h
e rover
s
are coor
di
nate
d
we
ll
. However, rovers cannot
b
e eas
il
yte
l
e-operate
d
s
i
nce t
his
∗
W
et
h
an
k
Apurva Mu
dg
a
lf
or
hi
s
h
e
l
p. T
hi
s researc
h
was part
ly
supporte
dby
NSF awar
d
sun
d
er contract
s
I
TR/AP0113881
,
IIS-0098807
,
and IIS-0350584. The views and conclusions contained in this document
are t
h
ose o
f
t
h
e aut
h
ors an
d
s
h
ou
ld
not
b
e
i
nterprete
d
as represent
i
n
g
t
h
eo
ffi
c
i
a
l
po
li
c
i
es, e
i
t
h
er expresse
d
or
i
mp
li
e
d
,o
f
t
h
e sponsor
i
n
g
or
g
an
i
zat
i
ons, a
g
enc
i
es, compan
i
es or t
h
e U.S.
g
overnment
.
3
L.E. Parker et al. (eds.)
,
M
ulti-Robot Systems. From Swarms to Intelligent Automata. Volume III
,
3
–
14
.
c
2005 Springer. Printed in the Netherlands
.
4
T
ove
y
, et al.
r
equ
i
res a
l
arge num
b
er o
fh
uman operators an
di
s commun
i
cat
i
on
i
ntens
i
ve
,
e
rror prone, an
d
s
l
ow. Ne
i
t
h
er can t
h
ey
b
e
f
u
ll
y preprogramme
d
s
i
nce t
h
e
ir
a
ct
i
v
i
t
i
es
d
epen
d
on t
h
e
i
r
di
scover
i
es. T
h
us, one nee
d
stoen
d
ow t
h
em w
i
t
h
t
he
c
apa
bili
ty to coor
di
nate autonomous
l
yw
i
t
h
eac
h
ot
h
er. Cons
id
er,
f
or exam
-
pl
e, a mu
l
t
i
-ro
b
ot exp
l
orat
i
on tas
k
w
h
ere a team o
fl
unar rovers
h
as to v
i
s
i
t
a
num
b
er o
f
g
i
ven target
l
ocat
i
ons to co
ll
ect roc
k
samp
l
es. Eac
h
target must
be
vi
s
i
te
db
yat
l
east one rover. T
h
e rovers
fi
rst a
ll
ocate t
h
e targets to t
h
emse
l
ves,
a
n
d
eac
h
rover t
h
en v
i
s
i
ts t
h
e targets t
h
at are a
ll
ocate
d
to
i
t. T
h
e rovers
k
no
w
t
h
e
i
r current
l
ocat
i
on at a
ll
t
i
mes
b
ut m
i
g
h
t
i
n
i
t
i
a
ll
y not
k
now w
h
ere o
b
stac
l
e
s
a
re
i
nt
h
e terra
i
n. It can t
h
ere
f
ore
b
e
b
ene
fi
c
i
a
lf
or t
h
e rovers to re-a
ll
ocate t
he
targets to t
h
emse
l
vesast
h
ey
di
scover more a
b
out t
h
e terra
i
n
d
ur
i
ng execut
i
on
,
f
or examp
l
e, w
h
enarover
di
scovers t
h
at
i
t
i
s separate
dby
a
big
crater
f
ro
m
i
ts next tar
g
et. S
i
m
il
ar mu
l
t
i
-ro
b
ot exp
l
orat
i
on tas
k
sar
i
se
f
or m
i
ne sweep
i
n
g,
searc
h
an
d
rescue operat
i
ons, po
li
ce operat
i
ons, an
dh
azar
d
ous mater
i
a
l
c
l
ean
-
i
n
g
, amon
g
ot
h
ers
.
Multi-robot coordination tasks are t
y
picall
y
solved with heuristic method
s
since optimizin
g
the performance is often computationall
y
intractable. The
y
a
re often solved with decentralized methods since centralized methods lack ro
-
bustness: if the central controller fails, so does the entire robot team. Marke
t
mechanisms, such as auctions, are
p
o
p
ular decentralized and heuristic multi
-
r
obot coordination methods (Rabideau et al., 2000). In this case, the robot
s
a
re the bidders and the tar
g
ets are the
g
oods up for auction. Ever
y
robot bid
s
o
n tar
g
ets and then visits all tar
g
ets that it wins. As the robots discover mor
e
a
bout the terrain durin
g
execution, the
y
run additional auctions to chan
g
eth
e
a
llocation of tar
g
ets to themselves. The resultin
g
auction-based coordinatio
n
s
y
stem is efficient in terms of communication (robots communicate onl
y
nu
-
meric bids) and com
p
utation (robots com
p
ute their bids in
p
arallel). It is there
-
f
ore not surprisin
g
that auctions have been shown to be effective multi-robot
c
oordination methods (Gerkey and Matar
i
´
c, 2002, Zlot et al., 2002, Thayer
´
e
t al., 2000, Goldberg et al., 2003). However, there are currently no systemati
c
methods for deriving appropriate bidding rules for given team objectives. I
n
t
hi
s paper, we propose t
h
e
fi
rst suc
h
met
h
o
d
an
dd
emonstrate
i
t
b
y
d
er
i
v
i
n
g
biddi
ng ru
l
es
f
or t
h
ree poss
ibl
e team o
bj
ect
i
ves o
f
t
h
emu
l
t
i
-ro
b
ot exp
l
orat
i
o
n
tas
k
.We
d
emonstrate exper
i
menta
ll
yt
h
at t
h
e resu
l
t
i
ng
biddi
ng ru
l
es
i
n
d
ee
d
e
x
hibi
t goo
d
per
f
ormance
f
or t
h
e
i
r respect
i
ve team o
bj
ect
i
ves an
d
compare
f
a
-
v
ora
bl
ytot
h
e opt
i
ma
l
per
f
ormance. Our researc
h
t
h
us a
ll
ows t
h
e
d
es
i
gners o
f
a
uct
i
on-
b
ase
d
coor
di
nat
i
on systems to
f
ocus on
d
eve
l
op
i
ng appropr
i
ate tea
m
obj
ect
i
ves,
f
or w
hi
c
h
goo
d biddi
ng ru
l
es can t
h
en
b
e
d
er
i
ve
d
automat
i
ca
ll
y
.
[...]... encounter situations that require agents to cooperate and perform a task In such situations it is often beneficial to assign a group of agents to a task, such as when a single agent cannot perform the tasks This paper investigates allocating tasks to disjoint robot teams, referred to as 15 L.E Parker et al (eds.), Multi-Robot Systems From Swarms to Intelligent Automata Volume III, 15–26 c 2005 Springer Printed... schedule is not available in advance, and 3 robots need to physically visit task locations to accomplish task completion (e.g., to push an object) Our approach to OMRTA relies on a computational and sensing fabric of networked sensors embedded into the 27 L.E Parker et al (eds.), Multi-Robot Systems From Swarms to Intelligent Automata Volume III, 27–38 c 2005 Springer Printed in the Netherlands 28... environments onto eight-connected uniform grids of size 51 × 51 and computed all costs between locations as the shortest distances on the grid Our auction-based coordination system used these costs to find an 10 Tovey, et al allocation of targets to robots and a path for each robot that visits all targets allocated to it We interfaced it to the popular Player/Stage robot simulator (Gerkey et al., 2003) to execute... applicability to the multiple-robot domain This work aims to correct that discrepancy by unearthing issues that arise while attempting to tailor these algorithms to the multiple-robot domain A well-known multipleagent coalition formation algorithm has been studied in order to identify the necessary modifications to facilitate its application to the multiple-robot domain Keywords: Coalition formation, fault-tolerance,... considerable distances from one another so that the best solution is to dispatch a robot team to each designated task area and hope that the team can autonomously complete the task The robots must then coalesce into teams responsible for each task The focus of this work is to investigate the various issues that arise while attempting to form multiple-robot coalitions using existing multi-agent coalition... communicate with each other and are aware of all tasks to be performed Each agent has a vector of real non-negative capabilities Bi =< bi , bi , bi > Each capability quantifies the ability to perform an r 1 2 action In order to assess coalitions and task execution, an evaluation function is attached to each capability type that transforms capability units into monet tary units It is assumed that there is a... fault tolerance was demonstrated Further algorithm modifications will permit more complex task execution by utilizing a MURDOCH (Gerkey and Mataric, 2002) style task allocation scheme within coalitions A future goal is to investigate methods of forming coalitions within a dynamic real-time environment The long-term goal is to develop a highly adaptive, fault tolerant system that would be able to flexibly... the multi-robot exploration task is to find an allocation of targets to robots and a path for each robot that visits all targets allocated to it so that the team objective is achieved Note that the robots are not required to return to their initial locations In this paper, we study three team objectives: M INI S UM: Minimize the sum of the path costs over all robots 6 Tovey, et al M INI M AX: Minimize... Therefore coalitions with one or more dominating members (resource contributors) tend to be heavily dependent on those members for task execution These dominating members then become indispensable Such coalitions should be avoided in order to improve fault tolerance Over reliance on dominating members can cause task execution to fail or considerably degrade If a robot is not a dominating member then... coalition with respect to a particular task can be calculated as follows BC = r1 × r2 × rn [ taskvalue ]n n (3) 22 Vig and Adams A perfectly balanced coalition has a coefficient of 1 The question is how to incorporate the balance coefficient into the algorithm in order to select better coalitions As previously discussed two cases arise: 1 Sufficient number of robots and high fault tolerance: Initially . Multi-Robot S
y
stems. From Swarms to Intelli
g
ent Automata
V
olume III
Volume III
Proceedings from the 2005 International Workshop
on.
From
Swarms to
Intelligent
Automata
ALAN C. SCHULTZ
N
av
y
Center
f
or Applied Research in A.I.
,
N
ava
l
Researc
h
La
b
oratory
,
Was
h
ington, DC,
Ngày đăng: 14/03/2014, 14:20
Xem thêm: Swarms to Intelligent Automata Volume III doc, Swarms to Intelligent Automata Volume III doc