ICRA-03 Tutorial on Ant-Based Mobile Robots: Robust ...
66
ICRA03 Tutorial http://www.cc.gatech.edu/fac/Sven.Koenig/icra03-tutorial.html ICRA-03 Tutorial on Ant-Based Mobile Robots: Robust Navigation and Coverage with Single Robots and Robot Teams Abstract Ant robots are simple and cheap robots with limited sensing and computational capabilities. This makes it feasible to deploy teams of ant robots and take advantage of the resulting fault tolerance and parallelism. Ant robots cannot use conventional planning methods due to their limited sensing and computational capabilities. Rather, their behavior is driven by local interactions. For example, ant robots can communicate via markings that they leave in the terrain, similar to ants that lay and follow pheromone trails. In the past couple of years, researchers have developed ant robot hardware and software and demonstrated, both in simulation and on physical robots, that single ant robots or teams of ant robots solve robot-navigation tasks (such as path following and terrain coverage) robustly. Thus, ant robots might provide a promising alternative to the currently very popular probabilistic robot navigation approaches. Researchers have also developed a theoretical foundation for ant robotics, based on ideas from real-time heuristic search, stochastic analysis, and graph theory. This half day tutorial on the current state of the art in ant robotics is given by experts who will give an overview of this exciting area of mobile robotics without assuming any prior knowledge on the topic. It will cover all important aspects of ant robotics, from theoretical foundations to videos of implemented systems. To this end, it will bring together the various research directions for the first time, including theoretical foundations, ant robot hardware, and ant robot software. Its primary objective is to give non-specialists a comprehensive overview of the state of the art of the field, for example, to allow researchers and students to do research in the area and to allow practitioners to evaluate the current state of the art in ant robotics. Consequently, it is of interest to anyone who is interested in mobile robotics and artificial intelligence. Speakers The material for the tutorial will be provided by the five researchers listed below, all of which are experts on the topic of ant robotics. They have been selected to cover the theory of ant robotics as well as practical approaches, including ant robot software and hardware. Sven Koenig (Georgia Institute of Technology, USA) Sven Koenig became an assistant professor in the College of Computing at Georgia Institute of Technology after receiving his Ph.D. in computer science from Carnegie Mellon University. He is the recipient of an NSF CAREER award, an IBM Faculty Partnership Award, the Raytheon Faculty Fellowship Award from Georgia Tech, and a Fulbright Fellowship Award. He is currently on the editorial board of the Journal of Artificial Intelligence Research (JAIR) and will co-chair the International Conference on Automated Planning and Scheduling (ICAPS) in 2004. Sven’s research centers around techniques for decision making (planning and learning) that enable situated agents to act intelligently in their environments and exhibit goal-directed behavior in real-time, even if they have only incomplete knowledge of their environment, limited or noisy perception, imperfect abilities to manipulate it, or insufficient reasoning speed. He has published over 60 papers on this topic in the artificial intelligence and robotics literature.
Transcript of ICRA-03 Tutorial on Ant-Based Mobile Robots: Robust ...
ICRA03 Tutorial http://www.cc.gatech.edu/fac/Sven.Koenig/icra03-tutorial.html
ICRA-03 Tutorial on Ant-Based Mobile Robots: RobustNavigation and Coverage with Single Robots and Robot
Teams
AbstractAnt robots are simple and cheap robots with limited sensing and computational capabilities. This makes it feasible todeploy teams of ant robots and take advantage of the resulting fault tolerance and parallelism. Ant robots cannot useconventional planning methods due to their limited sensing and computational capabilities. Rather, their behavior is drivenby local interactions. For example, ant robots can communicate via markings that they leave in the terrain, similar to antsthat lay and follow pheromone trails. In the past couple of years, researchers have developed ant robot hardware andsoftware and demonstrated, both in simulation and on physical robots, that single ant robots or teams of ant robots solverobot-navigation tasks (such as path following and terrain coverage) robustly. Thus, ant robots might provide a promisingalternative to the currently very popular probabilistic robot navigation approaches. Researchers have also developed atheoretical foundation for ant robotics, based on ideas from real-time heuristic search, stochastic analysis, and graphtheory. This half day tutorial on the current state of the art in ant robotics is given by experts who will give an overview ofthis exciting area of mobile robotics without assuming any prior knowledge on the topic. It will cover all important aspectsof ant robotics, from theoretical foundations to videos of implemented systems. To this end, it will bring together thevarious research directions for the first time, including theoretical foundations, ant robot hardware, and ant robot software.Its primary objective is to give non-specialists a comprehensive overview of the state of the art of the field, for example, toallow researchers and students to do research in the area and to allow practitioners to evaluate the current state of the art inant robotics. Consequently, it is of interest to anyone who is interested in mobile robotics and artificial intelligence.
SpeakersThe material for the tutorial will be provided by the five researchers listed below, all of which are experts on the topic ofant robotics. They have been selected to cover the theory of ant robotics as well as practical approaches, including antrobot software and hardware.
Sven Koenig (Georgia Institute of Technology, USA)
Sven Koenig became an assistant professor in the College of Computing at Georgia Institute ofTechnology after receiving his Ph.D. in computer science from Carnegie Mellon University. He is therecipient of an NSF CAREER award, an IBM Faculty Partnership Award, the Raytheon FacultyFellowship Award from Georgia Tech, and a Fulbright Fellowship Award. He is currently on theeditorial board of the Journal of Artificial Intelligence Research (JAIR) and will co-chair theInternational Conference on Automated Planning and Scheduling (ICAPS) in 2004. Sven’s researchcenters around techniques for decision making (planning and learning) that enable situated agents to actintelligently in their environments and exhibit goal-directed behavior in real-time, even if they haveonly incomplete knowledge of their environment, limited or noisy perception, imperfect abilities tomanipulate it, or insufficient reasoning speed. He has published over 60 papers on this topic in the
artificial intelligence and robotics literature.
ICRA03 Tutorial http://www.cc.gatech.edu/fac/Sven.Koenig/icra03-tutorial.html
Israel Wagner (Technion and IBM Haifa Research Lab, Israel)
Israel Wagner received his B.Sc. degree in computer engineering from the Technion (Israel Institute ofTechnology), Haifa, in 1987, cum laude, an M.Sc. degree in computer science from Hebrew University,Jerusalem, in 1990, cum laude, and a Ph.D. degree in computer science from the Technion in 1999. Hewas a research engineer at General Microwave, Jerusalem, from 1987 until 1990, when he joined theIBM Haifa Laboratories as a member of the technical staff. Israel is currently an adjunct senior lecturerin the Computer Science Department at the Technion. His research interests include multi-agentrobotics, manual and automatic VLSI design, computational geometry, and graph theory. He haspublished over 25 papers on these topics in the artificial intelligence, robotics, and VLSI literature andco-edited a special issue of the Annals of Mathematics and Artificial Intelligence on ant robotics in2001.
Andrew Russell (Monash University, Australia)
Andrew (Andy) Russell received his B. Eng. and Ph.D. degrees in 1972 and 1976, respectively, from theUniversity of Liverpool, in the U.K. After a brief period in the computer industry he took a position asEngineer working on robot design and applications in the Department of Artificial Intelligence at theUniversity of Edinburgh, Scotland. For the past 19 years he has been a member of academic staff at theUniversity of Wollongong and now at Monash University (both in Australia). Andy’s research interestsinclude robot tactile sensing, the design of robotic mechanisms and olfactory sensing for robots. He haswritten books on robot tactile sensing and odour sensing for mobile robots and published over 70refereed conference papers and journal articles describing his work in intelligent robotics.
Andrew will give a video presentation.
David Payton (HRL Laboratories, USA)
David Payton is principal research scientist and manager of the Cooperative and Distributed SystemsDepartment at HRL Laboratories in Malibu, California. He received his B.S. degree from UCLA in1979 and his M.S. degree from MIT in 1981. David is currently principal investigator for the DARPAPheromone Robotics project and is also involved in development of agent-assisted multi-usercollaboration tools. After joining HRL Laboratories in 1982, David has been involved in numerousprojects for the development of intelligent autonomous agents. This includes work on the DARPAAutonomous Land Vehicle project, the Unmanned Ground Vehicle and the development ofbehavior-based robot control. David has over 25 publications in the area and holds six patents.
Richard Vaughan (HRL Laboratories, USA)
Richard Vaughan received a D.Phil. in Computation from Oxford University in 1999, and a B.A.(Hons.) in Computing with Artificial Intelligence from Sussex University in 1993. He was apostdoctoral research associate at the Robotics Laboratory of the University of Southern California from1998 to 2001, and is currently a member of the technical staff at HRL Laboratories. Richard’s researchconcerns the mechanisms of intelligent behavior in individuals and groups of people, animals, robotsand software agents. His projects emphasize dynamic autonomy, scalability, interaction and simplicity.The majority of his more than 20 publications concern the functional analysis and recreation of aspectsof animal behavior, a methodology he calls "constructive ethology". His other papers are on simulation,interfacing and networking for robotics applications. Richard is a founding member of the Player/Stageproject, which creates tools for robotics and sensor network research.
Sven KoenigGeorgia Institute of TechnologyUSA
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
31
Sve
n K
oeni
g
Col
lege
of C
ompu
ting
Geo
rgia
Inst
itute
of T
echn
olog
y
Ant
Rob
otic
s
Terr
ain
Cov
erag
e
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
32
join
t wor
k w
ith:
Ove
rvie
w
- M
otiv
atio
n
- E
mpi
rical
Res
ults
- S
imul
atio
n-
Act
ual R
obot
s
- T
heor
etic
al R
esul
ts
One
-Tim
e or
Rep
eate
d C
over
age
of K
now
n or
Unk
now
n Te
rrai
n
- M
ine
Sw
eepi
ng-
Sur
veill
ance
- S
urfa
ce In
spec
tion
- G
uard
ing
Terr
ain
Str
uctu
re:
Ove
rvie
w:
- R
eal-T
ime
Sea
rch
Jona
s S
venn
ebrin
g, B
oles
law
Szy
man
ski (
RP
I), a
nd Y
axin
Liu
with
Sin
gle
Ant
Rob
ots
orTea
ms
of A
nt R
obot
s for
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
33
Mot
ivat
ion
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
34
chea
pgr
oups
of r
obot
s
limite
d co
mpu
tatio
nal a
nd s
ensi
ng a
bilit
ies
faul
t tol
eran
ce
para
llelis
m
DC
06
Rob
omow
Cye
Koa
la
Mot
ivat
ion
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
35
Mot
ivat
ion
You
wan
t to
build
a te
am o
f rob
ots
that
cov
er te
rrai
n re
peat
-ed
ly, f
or e
xam
ple,
to g
uard
a m
useu
m a
t nig
ht.
The
terr
ain
coul
d be
initi
ally
unk
now
n.T
he te
rrai
n co
uld
chan
ge d
ynam
ical
ly.
The
rob
ots
have
ver
y no
isy
actu
ator
s or
sen
sors
.T
he r
obot
s ca
n fa
il.
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
36
App
roac
h 1:
PO
MD
P-B
ased
Nav
igat
ion
Arc
hite
ctur
e
You
wan
t to
build
a te
am o
f rob
ots
that
cov
er te
rrai
n re
peat
-ed
ly, f
or e
xam
ple,
to g
uard
a m
useu
m a
t nig
ht.
The
terr
ain
coul
d be
initi
ally
unk
now
n.T
he te
rrai
n co
uld
chan
ge d
ynam
ical
ly.
The
rob
ots
have
ver
y no
isy
actu
ator
s or
sen
sors
.T
he r
obot
s ca
n fa
il.
Pro
babi
listic
Pla
nnin
g
R. S
imm
ons
and
S. K
oeni
g. P
roba
bilis
tic R
obot
Nav
igat
ion
in P
artia
lly O
bser
vabl
e E
nvi-
ronm
ents
. In
Pro
ceed
ings
of t
he In
tern
atio
nal J
oint
Con
fere
nce
on A
rtifi
cial
Inte
llige
nce,
1080
-108
7, 1
995.
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
37
App
roac
h 1:
PO
MD
P-B
ased
Nav
igat
ion
Arc
hite
ctur
e
sim
ulat
orin
terf
ace
to X
avie
r
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
38
- ex
plic
itly
mod
els
all u
ncer
tain
ty u
sing
pro
babi
litie
s-
mai
ntai
ns a
rbitr
ary
prob
abili
ty d
istr
ibut
ions
ove
r th
e po
ses
- ut
ilize
s al
l ava
ilabl
e se
nsor
dat
a (la
ndm
arks
, rob
ot m
ovem
ents
)-
robu
st to
war
ds s
enso
r er
rors
(no
exp
licit
exc
eptio
n ha
ndlin
g re
quire
d)
Adv
anta
ges
of N
avig
atio
n w
ith P
OM
DP
s
- un
iform
, the
oret
ical
ly g
roun
ded
fram
ewor
k
App
roac
h 1:
PO
MD
P-B
ased
Nav
igat
ion
Arc
hite
ctur
e
Xav
ier
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
39
App
roac
h 2:
Ant
Rob
otic
s
You
wan
t to
build
a te
am o
f rob
ots
that
cov
er te
rrai
n re
peat
-ed
ly, f
or e
xam
ple,
to g
uard
a m
useu
m a
t nig
ht.
The
terr
ain
coul
d be
initi
ally
unk
now
n.T
he te
rrai
n co
uld
chan
ge d
ynam
ical
ly.
The
rob
ots
have
ver
y no
isy
actu
ator
s or
sen
sors
.T
he r
obot
s ca
n fa
il.
Pro
babi
listic
Pla
nnin
g
no lo
catio
n es
timat
es!
no p
lann
ing!
no d
irect
com
mun
icat
ion!
sim
pler
har
dwar
e an
d so
ftwar
e!A
nt R
obot
ics:
Ter
rain
Cov
erag
e; S
ven
Koe
nig;
Geo
rgia
Inst
itute
of T
echn
olog
y; 2
003
10
- sh
ort l
ived
trac
es-
alco
hol [
Sha
rpe
et a
l.]-
heat
[Rus
sell]
- od
or [R
usse
ll el
at.]
- vi
rtua
l tra
ces
[Vau
ghan
et a
l.; P
ayto
n et
al.]
App
roac
h 2:
Ant
Rob
otic
s
Pat
h F
ollo
win
g us
ing
Phe
rom
one
Trac
es
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
311
- lo
ng li
ved
trac
es [S
venn
ebrin
g an
d K
oeni
g]
App
roac
h 2:
Ant
Rob
otic
s
Exp
lora
tion
and
Cov
erag
e us
ing
Phe
rom
one
Trac
es
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
312
The
oret
ical
Res
ults
(Rea
l-Tim
e S
earc
h)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
313
44 3
3 44
54 5
44 3
3 44
54 5
The
oret
ical
Res
ults
(R
eal-T
ime
Sea
rch)
u(s
):=
1 +
u(s
)
Initi
ally
, the
u-v
alue
su(s
)are
zer
o fo
r al
l sta
tess.
1.s
:= s
tart
loca
tion
2.s’
:= a
nei
ghbo
ring
loca
tion
ofs w
ith a
min
imal
u-v
alue
3. 4. m
ove
the
ant r
obot
to lo
catio
ns’5.
go
to 2
.
Nod
e C
ount
ing
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
314
00
00
0
00
00
0
00
00
00
00
00
00
00
00
30
00
0
00
00
0
00
00
00
00
00
00
00
00
31
00
0
20
00
0
00
00
00
00
00
00
00
00
31
00
0
22
00
0
10
00
00
00
00
00
00
00
31
00
0
22
10
0
12
00
00
00
00
00
00
00
31
10
0
22
10
0
12
00
02
00
00
00
00
00
31
10
0
22
20
0
12
00
02
10
00
01
00
00
31
20
0
22
20
0
12
00
02
10
00
11
10
00
time
step
0tim
e st
ep 1
time
step
2tim
e st
ep 3
time
step
4tim
e st
ep 5
time
step
6tim
e st
ep 7
31
20
0
22
20
0
12
00
02
10
00
11
10
00
The
oret
ical
Res
ults
(R
eal-T
ime
Sea
rch)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
315
The
oret
ical
Res
ults
(R
eal-T
ime
Sea
rch)
prog
ram
min
g: J
onas
Sve
nneb
ring
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
316
The
num
bers
(=
mar
king
s) c
oord
inat
e th
e an
t rob
ots!
010
2030
4050
6070
8090
100
0
1000
2000
3000
4000
5000
6000
Cov
erag
e N
umbe
r
Cover Time (steps)
Nod
e C
ount
ing
Ran
dom
Wal
k
node
cou
ntin
g
rand
om w
alk
cover time
cove
rage
num
ber
The
oret
ical
Res
ults
(R
eal-T
ime
Sea
rch)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
317
05
1015
0
200
400
600
800
1000
1200
1400
Nod
e C
ount
ing:
Sha
red
Mar
king
s
Nod
e C
ount
ing:
Indi
vidu
al M
arki
ngs
Time (steps)
Num
ber
of R
obot
s
node
cou
ntin
g (in
divi
dual
mar
king
s)
node
cou
ntin
g (s
hare
d m
arki
ngs)
num
ber
of r
obot
s
The
num
bers
(=
mar
king
s) c
oord
inat
e th
e an
t rob
ots!
cover time
The
oret
ical
Res
ults
(R
eal-T
ime
Sea
rch)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
318
u(s
) :=
1 +
u(s
’)if
u(s
)≤ u
(s’)
then
u(s
) :=
1 +
u(s
)u
(s):
=m
ax(
1 +
u(s
), 1
+ u
(s’))
Kor
f’s L
RTA
* (=
Wag
ner’s
VA
W)
u(s
):=
1 +
u(s
)N
ode
Cou
ntin
g
Initi
ally
, the
u-v
alue
su(s
)are
zer
o fo
r al
l sta
tess.
1.s
:= s
tart
loca
tion
2.s’
:= a
nei
ghbo
ring
loca
tion
ofs w
ith a
min
imal
u-v
alue
3. 4. m
ove
the
ant r
obot
to lo
catio
ns’5.
go
to 2
.
or or or
44 3
3 44
54 5
44 3
3 44
54 5
Wag
ner’s
Upd
ate
Rul
eT
hrun
’s U
pdat
e R
ule
The
oret
ical
Res
ults
(R
eal-T
ime
Sea
rch)
Add
ition
al R
eal-T
ime
Sea
rch
Met
hods
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
319
The
orem
:
Team
s of
ant
rob
ots
that
all
use
the
sam
e re
al-t
ime
sear
ch m
etho
d co
ver
all s
tron
gly
conn
ecte
d gr
aphs
Pro
of:
The
oret
ical
Res
ults
(R
eal-T
ime
Sea
rch)
repe
ated
ly.
QE
D
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
320
The
oret
ical
Res
ults
(R
eal-T
ime
Sea
rch)
num
ber
of v
ertic
es
time to reach goal (log scale)
Kor
f’s L
RTA
*
Nod
e C
ount
ing
2#ver
tices
/2+
1/2 -3
cove
r tim
e gu
aran
teed
to b
e no
wor
se th
an O
(#ve
rtic
es d
iam
eter
)on
any
str
ongl
y co
nnec
ted
grap
h
10100
1000
1000
0
1000
00
1e+0
6
1e+0
7
1e+0
8
1e+0
9
1e+1
0
1e+1
1
1e+1
2 050
100
150
200
250
300
350
400
num
ber
of v
ertic
es (
n)
star
tgo
al
[Koe
nig
and
Sim
mon
s, 1
992]
[Koe
nig
and
Sim
mon
s, 1
992]
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
321
The
oret
ical
Res
ults
(R
eal-T
ime
Sea
rch)
num
ber
of v
ertic
es
time to reach goal (log scale)
Nod
e C
ount
ing
#ve
rtic
essq
rt((
1/6-
epsi
lon)
#ver
tices
)
10100
1000
1000
0
1000
00
1e+0
6
1e+0
7
1e+0
8
1e+0
9
1e+1
0
1e+1
1 050
100
150
200
250
300
350
400
num
ber
of v
ertic
es (
n)
sim
ulat
ion
form
ula
star
t
goal
Kor
f’s L
RTA
*co
ver
time
guar
ante
ed to
be
now
orse
than
O(#
vert
ices
dia
met
er)
on a
ny s
tron
gly
conn
ecte
d gr
aph
[Koe
nig
and
Sim
mon
s, 1
992]
[Koe
nig
and
Szy
man
ski,
1999
]
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
322
u(s
) :=
1 +
u(s
’)if
u(s
)≤ u
(s’)
then
u(s
) :=
1 +
u(s
)u
(s):
=m
ax(
1 +
u(s
), 1
+ u
(s’))
poly
nom
ial c
over
tim
epo
lyno
mia
l cov
er ti
me
poly
nom
ial c
over
tim
e
u(s
):=
1 +
u(s
)ex
pone
ntia
l cov
er ti
me
Initi
ally
, the
u-v
alue
su(s
)are
zer
o fo
r al
l sta
tess.
1.s
:= s
tart
loca
tion
2.s’
:= a
nei
ghbo
ring
loca
tion
ofs w
ith a
min
imal
u-v
alue
3. 4. m
ove
the
ant r
obot
to lo
catio
ns’5.
go
to 2
.
or or or
The
oret
ical
Res
ults
(R
eal-T
ime
Sea
rch)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
323
Is th
e w
orst
-cas
e co
ver
time
of a
sin
gle
ant r
obot
that
use
s no
de c
ount
ing
poly
nom
ial o
r ex
pone
ntia
lin
the
num
ber
of v
ertic
es (
an a
dver
sary
can
cho
ose
the
grap
h to
polo
gy, t
he s
tart
ver
tex,
the
goal
ver
tex,
and
the
tie-b
reak
ing
rule
),
- if
the
stro
ngly
con
nect
ed g
raph
s ar
e di
rect
ed?
- if
the
stro
ngly
con
nect
ed g
raph
s ar
e un
dire
cted
?-
if th
e st
rong
ly c
onne
cted
gra
phs
are
undi
rect
ed g
rids?
yes
(see
3 s
lides
ago
)ye
s (s
ee 2
slid
e2 a
go)un
know
n
The
oret
ical
Res
ults
(R
eal-T
ime
Sea
rch)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
324
The
oret
ical
Res
ults
(R
eal-T
ime
Sea
rch)
num
ber
of r
obot
s
cover time
Kor
f’s L
RTA
*N
ode
Cou
ntin
g
Wag
ner’s
Upd
ate
Rul
e
Thr
un’s
Upd
ate
Rul
e
[Koe
nig
et. a
l., 2
001]
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
325
The
oret
ical
Res
ults
(R
eal-T
ime
Sea
rch)
Nod
e C
ount
ing
Kor
f’s L
RTA
*
Wag
ner’s
Upd
ate
Rul
eT
hrun
’s U
pdat
e R
ule
[Koe
nig
et. a
l., 2
001]
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
326
Em
piric
al R
esul
ts(S
imul
atio
n an
d A
ctua
l Rob
ots)
BO
RG
Lab
[Sve
nneb
ring
and
Koe
nig,
200
3]
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
327
Em
piric
al R
esul
ts (
Act
ual R
obot
s)
Ant
rob
ots
that
all
use
node
cou
ntin
g ar
e ea
sy to
impl
emen
t!
40
4
40
4
43
2
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
328
Ant
Rob
ot H
ardw
are
A: t
rail
sens
orB
: tra
il se
nsor
C: p
enD
: mic
ro-c
ontr
olle
rE
: RS
232
inte
rfac
e
Tha
nks
to A
shw
in R
am fo
r th
e ha
rdw
are.
Em
piric
al R
esul
ts (
Act
ual R
obot
s)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
329
Ant
Rob
ot S
oftw
are
our
ant r
obot
s us
e a
sche
ma-
base
d na
viga
tion
stra
tegy
with
an
obst
acle
avo
idan
ce b
ehav
ior
and
a tr
ail-a
void
ance
beh
avio
r
Em
piric
al R
esul
ts (
Act
ual R
obot
s)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
330
our
ant r
obot
s us
e a
sche
ma-
base
d na
viga
tion
stra
tegy
with
an
obst
acle
avo
idan
ce b
ehav
ior
and
a tr
ail-a
void
ance
beh
avio
r
Ant
Rob
ot S
oftw
are
Em
piric
al R
esul
ts (
Act
ual R
obot
s)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
331
Em
piric
al R
esul
ts (
Act
ual R
obot
s)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
332
Our
ant
rob
ots
cove
r cl
osed
terr
ain
even
if-
they
don
’t kn
ow th
e te
rrai
n in
adv
ance
or
the
terr
ain
chan
ges,
- so
me
ant r
obot
s fa
il,-
som
e an
t rob
ots
are
mov
ed w
ithou
t rea
lizin
g th
is, o
r-
som
e tr
ails
are
des
troy
ed.
dest
roye
d ar
eas
of tr
ails
low
-inte
nsity
trai
ls
Em
piric
al R
esul
ts (
Sim
ulat
ion
and
Act
ual R
obot
s)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
333
end
of fi
rst c
over
age
end
of th
ird c
over
age
The
terr
ain
gets
sat
urat
ed w
ith tr
ails
ove
r tim
e.
Em
piric
al R
esul
ts (
Act
ual R
obot
s)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
334
40
4
40
4
43
2
rand
omly
dro
p a
drop
of in
k in
to th
is c
ell
incr
ease
this
num
ber
by o
new
ith p
roba
bilit
y (1
6-4)
/16
with
pro
babi
lity
(170
-u(s
))/1
70 d
o:u(s
):=
1 +
u(s
)
Initi
ally
, the
u-v
alue
su(s
)are
zer
o fo
r al
l sta
tess.
1.s
:= s
tart
loca
tion
2.s’
:= a
nei
ghbo
ring
loca
tion
ofs w
ith a
min
imal
u-v
alue
3. 4. m
ove
the
ant r
obot
to lo
catio
ns’5.
go
to 2
.
Em
piric
al R
esul
ts (
Mod
elin
g w
ith R
eal-T
ime
Sea
rch)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
335
010
2030
4050
6070
8090
100
0
500
1000
1500
2000
2500
3000
Cov
erag
e N
umbe
r
Cover Time (minutes)
Mod
ified
Nod
e C
ount
ing T
eam
Bot
s S
imul
atio
n of
Peb
bles
Tea
mB
ots
Sim
ualti
on o
f Ran
dom
Wal
k
Cover Time (steps)
0
2000
0
1666
6
1333
3
1000
0
6666
3333
mod
ified
nod
e co
untin
g
robo
t(w
ithou
t tra
il re
mov
al)
cove
rage
num
ber
cover time
Em
piric
al R
esul
ts (
Sim
ulat
ion)
rand
om w
alk
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
336
trai
ls a
fter
first
cov
erag
etr
ails
afte
r te
nth
cove
rage
cleaningno cleaning
with
one
ant
rob
otw
ith o
ne a
nt r
obot
Em
piric
al R
esul
ts (
Sim
ulat
ion)
end
of fi
rst c
over
age
end
of te
nth
cove
rage
withwithouttrail removal trail removal
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
337
010
2030
4050
6070
8090
100
050100
150
200
250
300
350
400
450
500
Cover Time (minutes)
Cov
erag
e N
umbe
r
Tea
mB
ots
Sim
ulat
ion
of P
ebbl
es w
ith C
lean
ing
Tea
mB
ots
Sim
ulat
ion
of P
ebbl
es w
ithou
t Cle
anin
g
robo
t (w
ithou
t tra
il re
mov
al)
robo
t (w
ith tr
ail r
emov
al)
cove
rage
num
ber
cover time
Em
piric
al R
esul
ts (
Sim
ulat
ion)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
338
terr
ain
size
cover time
Em
piric
al R
esul
ts (
Sim
ulat
ion)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
339
num
ber
of a
nt r
obot
s
cover time
Em
piric
al R
esul
ts (
Sim
ulat
ion)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
340
25 x
25
met
ers
85 h
ours
with
out a
nyan
t rob
otge
tting
stu
ck
10 a
nt r
obot
s
Em
piric
al R
esul
ts (
Sim
ulat
ion)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
341
The
Fut
ure
smal
l inf
rare
d tr
ance
iver
s as
sm
art m
arke
rs(s
imila
r in
tere
stin
g w
ork
is p
erfo
rmed
at U
SC
and
oth
er in
stitu
tions
)
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
342
Sum
mar
y
Rea
l-tim
e se
arch
met
hods
pro
vide
an
inte
rest
ing
mea
ns fo
rco
ordi
natin
g si
ngle
ant
rob
ots
or te
ams
of a
nt r
obot
s th
atco
ver
know
n or
unk
now
n te
rrai
n on
ce o
r re
peat
edly
. The
yle
ave
mar
king
sin
the
terr
ain,
sim
ilar
tow
hats
ome
ants
do.
The
ant
rob
ots
robu
stly
cov
er te
rrai
n ev
en if
the
robo
ts a
rem
oved
with
out r
ealiz
ing
this
, som
e ro
bots
fail,
and
som
em
arki
ngs
get d
estr
oyed
. The
rob
ots
do n
ot e
ven
need
to b
elo
caliz
ed.
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
343
Sel
ecte
d P
ublic
atio
nsA
nt R
obot
ics
[Sve
nneb
ring
and
Koe
nig,
2002
]J.S
venn
ebrin
gan
dS
.Koe
nig.
Bui
ldin
gTe
rrai
n-C
over
ing
Ant
Rob
ots.
Tech
nica
l Rep
ort,
GIT
-CO
GS
CI-
2002
/10,
Col
lege
of C
ompu
ting,
Geo
rgia
Inst
itute
of T
echn
olog
y,A
tlant
a (G
eorg
ia),
200
2
[Koe
nig
et. a
l., 2
001]
S. K
oeni
g, B
. Szy
man
ski a
nd Y
. Liu
. Effi
cien
t and
Inef
ficie
nt A
nt C
over
age
Met
hods
. Ann
als
of M
athe
mat
ics
and
Art
ifici
al In
telli
genc
e, 3
1, 4
1-76
, 200
1.
[Sve
nneb
ring
and
Koe
nig,
2003
]J.S
venn
ebrin
gan
dS
.Koe
nig.
Tra
il-La
ying
Rob
ots
forR
obus
tTer
rain
Cov
erag
e. In
Pro
ceed
ings
of t
he In
tern
atio
nal C
onfe
renc
e on
Rob
otic
s an
d A
utom
atio
n, 2
003.
Rea
l-Tim
e S
earc
h
[Koe
nig
and
Szy
man
ski,
1999
] S. K
oeni
g an
d B
. Szy
man
ski.
Valu
e-U
pdat
e R
ules
for
Rea
l-Tim
eS
earc
h. In
Pro
ceed
ings
of t
he N
atio
nal C
onfe
renc
e on
Art
ifici
al In
telli
genc
e, 7
18-7
24, 1
999.
[Koe
nig
and
Sim
mon
s, 1
996a
] S. K
oeni
g an
d R
.G. S
imm
ons.
Eas
y an
d H
ard
Test
beds
for
Rea
l-Tim
eS
earc
h A
lgor
ithm
s. In
Pro
ceed
ings
of t
he N
atio
nal C
onfe
renc
e on
Art
ifici
al In
telli
genc
e, 2
79-2
85,
1996
.
[Koe
nig
and
Sim
mon
s, 1
996b
] S. K
oeni
g an
d R
.G. S
imm
ons.
The
Influ
ence
of D
omai
n P
rope
rtie
s on
the
Per
form
ance
of R
eal-T
ime
Sea
rch
Alg
orith
ms.
Tec
hnic
al R
epor
t, C
MU
-CS
-96-
115,
Sch
ool o
fC
ompu
ter
Sci
ence
, Car
negi
e M
ello
n U
nive
rsity
, Pitt
sbur
gh (
Pen
nsyl
vani
a), 1
996.
[Koe
nig
and
Sim
mon
s, 1
995]
S. K
oeni
g an
d R
.G. S
imm
ons.
Rea
l-Tim
e S
earc
h in
Non
-Det
erm
inis
ticD
omai
ns. I
n P
roce
edin
gs o
f the
Inte
rnat
iona
l Joi
nt C
onfe
renc
e on
Art
ifici
al In
telli
genc
e, 1
660-
1667
,19
95.
[Koe
nig
and
Sim
mon
s, 1
992]
S. K
oeni
g an
d R
.G. S
imm
ons.
Com
plex
ity A
naly
sis
of R
eal-T
ime
Rei
n-fo
rcem
ent L
earn
ing
App
lied
to F
indi
ng S
hort
est P
aths
in D
eter
min
istic
Dom
ains
. Tec
hnic
al R
epor
t,C
MU
-CS
-93-
106,
Com
pute
r S
cien
ce D
epar
tmen
t, C
arne
gie
Mel
lon
Uni
vers
ity, P
ittsb
urgh
(P
enns
ylva
-ni
a), 1
992.
Ant
Rob
otic
s: T
erra
in C
over
age;
Sve
n K
oeni
g; G
eorg
ia In
stitu
te o
f Tec
hnol
ogy;
200
344
Add
ition
al In
form
atio
n
http
://w
ww
.cc.
gate
ch.e
du/fa
c/S
ven.
Koe
nig/
ants
.htm
l
Israel WagnerTechnion andIBM Haifa Research LabIsrael
�������������� ��
������� ���������������� !���������"�!�#$��"�!�#���%&!������
' ( ) * + ,.-0/ 12* 3546+ )
7 869;:�< = > <�? < @ AB C D
7 E F G H H = I G E F J K A F G LMA5? < @N6O L�P Q F G R S J T = G E T G6U�G P < R F L�G E F V W G T X E = O E
Y Z Z [�\ ] ] ^6^6^�_ ` a _ Z b ` Y C c d C _ B ` _ c e ] f ^ B g C b h ]
i j k l m n m k o p q r s q t u v w i x y u z j v t { s | }�~6� v p � j o m j { � � � � �
�����.� ������� ���
�&��� ( + ) �6* � � �64�� * 46� (�* � �6� + �6+�� ���6) + ( ( � �6+�) + ( �6, � ( �6+ ( �6+ � � * , , ) � � �6( � 46+ ( ( �6¡6+ ( �6� � +�� 46�6+ ) + 46��, � ��� � * � � �646(�* 46¡£¢ ) + ¤6�6+ 46�.¢ * � , �6) + (�&¥ �6+ ( � � �646(.�¦�§;¨ © ª6«�© ¬ ®�©�¯ ° ± ² ³�± ´.® µ «�³ ¶ · ¶ µ «�µ ª ¸�¹ ° © ª ² ° ®«�² ¹ ¨�«�± ° � ´ ´ µ ¹ µ º ª5ª ¨ © º�© º »�® µ º ¯ ¶ �± º �¼
¦�½�© º�¾��¿ ± ° ° ± ¾0µ ¸ © ®�´ ± °.° ± ¿ ± ª µ ¹ ®M¼�&� �6)M* ( ( �6���6� � �64��ÀÁ�6+�* 46�  (�( � ) + 4636� ��� (�) + , * � + ¡�� �£� � (�Ã�+ * Ä646+ ( (��, � ��� � + ¡�( + 46( � 4636�6� �6���6�6� � 463�* 46¡£� �6�����646� � * � � �64£� ���6, � + (��
Å6Æ£Ç È Å É�Ê Ë�Ì6È Í�Æ Î�È Å Î Æ6Ï Ð�Ñ6Í È Æ6Å�È Ç�Ò Ï È Í È Ò Ñ6Ê Ê Ó;È Ð�Ô�Æ6Ï Í Ñ6Å6Í� ' 4£� �6+�( + ¤6�6+ , �Á( �6��+�+ Õ�* ���6, + (��6¢M* 46* , 5( � ( �6( � ���6, * � � �64�* 46¡, * � + Õ6�6+ ) � ��+ 46� ( �6¡6+ ���646( � ) * � � 463�� �6+�) � � �6( � 46+ ( (��6¢�* 46� Ö ) � � �6� (
i j k l m n m k o p q r s q t u v w i x y u z j v t { s | }�~6� v p � j o m j { � � � � ×
ØÚÙ ��Û � �0Ü
� -�46� (�* 46¡�Ý�) * �6�6(Þàß�á�â�ã�ä å æ.çÁä ä6á�è é æ�å ê ë£é ì í�ä6á�æ�î�ï é ð�ñ�äÞàò£æ�å ê ë�é ì í�ä6é æ�ó�ô5õ�ô ì ç�ä
ö�÷ ø.ù < E ú.F X G ÷ Q H G R T K T H Gö�û.ø.ù < E ú.F X G6:�< L�= H F O E�T K T H GÞàüMé å ï á�ì ì ý æ�î�é æ�ó�å ñ�ç�þ , * 46Ä�+ ��� � ��+
ÿ���� � ��� ��� �� � � � � � � � �Þ���ò��ê ü��Þ���é â�üMï á�� ç ô å ä �
ö 9 ø N < E ú � Nö N ø.ù�� N6O E F = E Q O Q A ø E F ù < H � �ÿ�� � �"! � � � � � � ��# � $�� � � % & � � � � � �' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' '( � � � � )*� + ,�-�� � ./ + $ $ 0�1 + � % , � � � � 2 3�� % # 4�� � $ � 5 # � �62 7 & � �"8 � �6� � 29 & # $ � �6� +�:�# � � ; � , � 2 <"� # �>=�� + $ � � 2 8�� � ��?� � $ # +
i j k l m n m k o p q r s q t u v w i x y u z j v t { s | }�~6� v p � j o m j { � � � � @
A �0� B�C ��Û D>E ���GFIH��*J>K>EL � < R < ú = I L�M
Þàë£é ì í�é ï á�ã�æ�ó�å ñ�ç�î�ï é ð�ñÞàì ç é N�ç�å ï é ô ç�äÞàæ�é N�ý î�é å ç�ã�ä ý æ�î�å ñ�ç�å ï é ô ç�ä
LÚ÷ O < L�P H G M ÷ ú I G ø E F ù < H � � P�Q R �TS < U � U < U V ÷ ú I G N6O Q E F = E I W X M-�( ( �6���6� � �64�� � �6+ ) +�� (�*ZY 46+ Õ�� [��6�6� 46� + )M� 4�+ * � �£�6+ ) � + Õ6��6�6� 46� � 463�� �£�646+��6¢M� �6+��6�6� 35�6� 4�3�+ ¡636+ (\T]_^ ß�ã�ì ç"�
` a�b c d e d f g d h f e d i c j k i l f g g m n l e d o n p�q6d c�r�s p n e�t u v w x q y{z |"} ~� a�b e n l f e d i c p j � n d c ��f e m n l e n ��q � i
� � J G F b.\ � C b � Z � � � �� � J G F C b � Z � � �M\ �T� C b � Z � � � �_� ��� d D�D b g � � � �� � � O < H O E I b �
b � e ��i�� i l � i l n � f � n c e p i o o � � �{e � n p f � n m n l e n � sg n e e � n �Tf o e p n � � n c e d f g g �{� ��p i � n f l � d e l f l ��i l � n l
i j k l m n m k o p q r s q t u v w i x y u z j v t { s | }�~6� v p � j o m j { � � � �
¡ A C¢B�ÜT£��¥¤¦J0Û Ü
§ ¨
©
ª ��
��
« # � �¬�
i j k l m n m k o p q r s q t u v w i x y u z j v t { s | }�~6� v p � j o m j { � � � � ®
¡ A C¢B�ÜT£��¥¤¦J0Û Ü
§ ¨
©
ª ��
��
« # � �¬¥ ¯ ° ± �² ¯ ± ¯ ° ± ³ ´
i j k l m n m k o p q r s q t u v w i x y u z j v t { s | }�~6� v p � j o m j { � � � � µ
¡ A C ���>¶ ��K0Ü·¡ Ù Û ÜTH·¸º¹*»�Û Ü
� -�( ( �6��+�* 4_¼Á¡636+�-�46�.12* , Ä£�64�*�35) * �6�¾½_¿ À�Á Â�Ã� ¼Á�6, + )�Ä� 5� , +ZÅ Æ6ÇÉÈ�Ê�*�� 5� , +�� 4£Ã��6� � ��+ * � �£+ ¡636+�� 4ZÂ2� (�6� ( � � + ¡ Ë Ë�Ñ6Ò Í Ê Ó �646� +
� ÀÁ�6+ �6) + �2�' ¢�Ì2� ("¼Á�6, + ) � * 4�Å Í Î Ï ± Ð Ñ Ò Ð Ð Ó Ô Õ ¬�Ö Ô × Ï ± Ð Ñ Ò Ð Ð Ó Ô Õ ¢ �6)M* , ,.�6+ ) � � � + ( Ô È �� �6+ 4�* 4¥Æ6ØÉÙÚ�6) �6� + ( (�� �646�6+ ) 36+ (�� ��* 4�¼��6, + )M� 5� , +�·Ú �6Ã�, �6463�¡6�6+ (�� �.� * Ä6+ZÛ
ÜIÝ¥Þ ß¥à á�â ã¥äTå æ�åZߥçÉè é6è â à ç_è ê�èë 1 # � � 2 7�; � 2 =�+ � 5 + ì 2 í�# 0 # +Éî ï ï ð ñ
ÜòÝ¥Þ ß¥ó"á�â{ã¥äTå æ�å6ó¥ç_ô õ ö ß¥Þ ÷�áë :�# � � ; � , � 2 )*# ì � + 2 1 + � % , � � � �"î ï ï ð ñ
i j k l m n m k o p q r s q t u v w i x y u z j v t { s | }�~6� v p � j o m j { � � � � ø
¡ A Cúùüûþý>ÿ ��� ���>¶Gû���� �������üû ý>ÿ ���� 1��6* ��� ¢�Ì�� (�46�646Ö ¼Á�6, + ) � * 4_Û� ¼ÁÕ6� + 46¡6+ ¡�¼��6, + )�Ä� 5� , +ZÅ ÆZÆ6ÇTÈ6�*�� 5� , +�� 4£Ã��6� � ��* , ,.+ ¡636+ (�� 46Â2* ) +��6� ( � � + ¡� �6+�( * ��+�46�6� � + )M�6¢M� � ��+ (
� ¼ÁÕ6* ���6, +��* Å6Æ6Å�� Â� Ê Ë Ï È Ñ6Å 36) * �6���
* 4�¼�¼�Ä��6¢M�6) ¡6+ )������ ����������������������� ���Å *�� 5� , +�Ã���+ ) +�+ * � �£+ ¡636+��6� � �6) (�+ Õ6* � � , Í ��È Ò Ë È
��
� "!$#§
¨ ©
ª
i j k l m n m k o p q r s q t u v w i x y u z j v t { s | }�~6� v p � j o m j { � � � � %
¡ A Cúùü¡ Ù Û ÜTH&���>¶G¡þ£���Ü��>¶0ÜT¶G¡ Ù Û ÜTH
� ÀÁ�6+ �6) + �2�ÀÁ�6+��646� ¢ � + ¡��6* � �£�6¢M*�35) �6�6�£�6¢'&�* 36+ 46� (* �6�6, � 4�3�� �6+TÆZØÉÙÚ) �6, +�+ �6+ 46� �6* , , �+ 46� + ) (
* 46 Ë�Í Ë Å�(6Ë ( Â� Ê Ë Ï*)6Ó�Ò Ê Ë Å �6¢M( �6��+��6) ¡6+ ) È�·Ú �6Ã�, �6463�¡6�6+ (�� �.� * Ä6+ZÛ+ ( � � ��* � � �64£Ã�� � �£( � ���6, * � � �54�� , + ( (�� �6* 4�+,� Ý��6+ ( (.�� � ��+�¡6+ � ) + * ( + (�Ã�� � ��&5� � �6�.46���6) �6�6¢M 5+ � /� -��6�6, � � * � � �64�� Ô�Ñ6Í Ï Æ6Ê Ê È Å É *�46+ � Ã��6) Ä6�6� �6* ��� (��� �646� � 46�6�6�6( , �+ Õ��6, �6) � 463�* , ,�+ ¡636+ ( �£* (- Å6È Î Æ6Ï Ð�Ê Ó * (��6�6( ( � � , +�/. + �.�6(�¢ � ) ( �.� �646( � ¡6+ )�*�( � ���6, + )M��+ � �6�6¡��6¢�Å ( � �6� �6* ( � � � È�6* � ) �6, , � 46310
i j k l m n m k o p q r s q t u v w i x y u z j v t { s | }�~6� v p � j o m j { � � � � 2 3
� -�4 Ç È Ð�Ô�Ê Ë�Ï Ñ6Å�(6Æ6Ð4��Ñ6Ê 5 �64¥Ì-6�798 :�;�85) �6, + ¡ �
� (�* 4 Ë Ï É�Æ�(6È Ò �6) �6� + ( (�� �6* �.+ �6+ 46� �6* , , �* ) ) � �6+ (�* �.* Ç Í Ñ6Í È Æ6Å6Ñ6Ï Ó(6È Ç Í Ï È Ì Í È Æ6Å�<<<<=
* 46¡£� �6+�� �646�6+ ) 36+ 46� +TÅ Y ��� Õ6� 463{[ ÈM) * � +�� � < � * 4 � +�+ Õ6�6) + ( ( + ¡� �� �6+ Ì6Ê Ñ6Å�5 Ë Í�Í È Ð�Ë > ë )*� � , & +�2 ?�� % , + �6# �"î ï @ ð ñ .
A�����H���Û Û � �BDC>¹��DE ���>¶0�*¤ C ��Û D
F GHGIG JJ KLMLMN O PLM QQ RST
UV W�XZY�[�\ ] [V \ X�^ _�_ \ `a bcd e�f g he�f i h f g hjf i hji�k glm nn
o poqr stu v�w x9y z {*| } y~������ � ����� ����� �
i j k l m n m k o p q r s q t u v w i x y u z j v t { s | }�~6� v p � j o m j { � � � � 2 2
A�����H���Û Û � �BDC>¹��DE ���>¶0�*¤ C ��Û D
ÿI1�& # � , � � � �6�î $ ì ; + � � � � ð .í� �� � + � � � � �"�# � -� � % 2 � � +�# � 0�� -��" $ ì ��� � �¾� ���"2
ÿü5 0�+ # � $ � � -�# & , .��� � � ¬���Ó ��Ó � Ö Ñ�� Ö Ñ Î Õ � Õ î � � % � $ ð
-� + Z�ò� ��� � � � % � $"% � ; +�� � �"�� �-�ë ?�# � 2 ?�� �Z2 4�� ; # � � 2 9 �"î / �����9� � � ð ñ Ö × Ð ¡��_% # ��5 �# ��& # + ì �# ��¢Z£ ¤ ¥ ¦ ë # ��� �"#�� Ö � � Í § Ö § ì + # � ñÿü5 0©¨�ªZ«�. �¬�® ¯�° î $ � + �6� � � � � � % ð
±²³ ²³µ´¶· ¸ ¹�¸ º ¹¼»· ¸ ¹�¸ º ¹ ½ ½
i j k l m n m k o p q r s q t u v w i x y u z j v t { s | }�~6� v p � j o m j { � � � � 2 �
¾ � ¤ Ù Û ���.� ���>E Bà� Ù Û ��� A B�Ü��0� A �0� C ��Û D
¿ À Á Â Ã Ä Ã Á Å Æ Ç È É Ç Ê Ë Ì Í ¿ Î Ï Ë Ð À Ì Ê Ñ É Ò ÓÕÔ�Ö Ì Æ × À Å Ã À Ñ Ø Ù Ù Ú Û Ü
Ý-Þ ßáàâ ã�ä�Þ å�æçéè1ê�â ã�æë1ì�äíä�Þ ßáì
î ï�ð ñ ò ó ô ñ õ�ö ð ñ ò�÷ ø ø�ò ó ù õ ö úZî û9ü ô ò ó ý þ ÿ ��� � þ�� ø�ð ñ ò � � � ñ ÿ ü�� � � �
� � ��� � ��� � � � � � � ����� � � ��� � � � � � � � � ��� ! � � � " ��� � � � � � � � � #$ " % � & � � � � �'� ( � � � ��� � � � � �)� � !�* � � � +
, - . / 0 1 0 . 2 3 4 5 6 4 7 8 9 : , ; < 8 = - 9 7 > 6 ? @�A�B 9 3 C - 2 0 - > D E E F G H
IKJMLONQPSRUT�VSWYXZNQ[]\ ^ T�_QPU`Kacb�^ W
dfehg iQj k l m�npo�q r k s�t uQvhw�xyg�r q)r k s�j n{zh|�j r |ps g r |~} � � � � �pj nQ��j �� j � j l s �{s ��g r l k qhm�n�r s
d�uQ�h����� m���k s i��� j � s npg � � g ��|�����m�s ��j l�r m�n�l g j npg n{eho��
� �'� � � � � � � � � � ��� � ¡ ��¢¤£ ¥ ¦ § ¨ © ª «�¬ ¥ ¨ § ®)§ £ ¯ ª�°�£ ¯ ± ¨ § ²³d�uQ�h��j ��´hn�m�zhn{l mp��s�µ{¶'· ¸ ¹�ºy»¤¼ ½ ¾ ½ �{|�s n¿r s�n�m�l�� m�k � g ��k s�j nl j iQs¤ÀÁ� Ã'Ä Å Æh��Ç�n�k s � �¶'Ȥµh¶
dfÉhÇ��'Ê�ˤ¾ · Ì Ë¤Í »¤Ì Î ½ ÏQg ����� m�g r |pt ÐhÑhÒÓ��Ô'ÕÖ×ÑSØ w��Ù � s ��s g l s ��k qhl � g � s � � s�l |�s � � g ��|���iQg � Ú�j n � l |�s � s � l j r s �Ù g k zhg q)� � mpl m{l |�s�n�s j � |���m¿��zhj l |{iQj n�j iQÇ�iÛl � g r s
dfo�g n{l |�j ��Ü � s � l s ��Ý�Þ k g � m���m�Þ�ß×ÑhÒàk s g �{l mpg n{eho��
, - . / 0 1 0 . 2 3 4 5 6 4 7 8 9 : , ; < 8 = - 9 7 > 6 ? @�A�B 9 3 C - 2 0 - > D E E F G á
IKJKLONcPSRUXâJKã
äpå�æ �cç�è é ê è ë�ì íŤê ì�îhï� ðyñ ò � s � l s �'ó ôõ öh÷cø ùûú ü �� � � ( � � % "�ý� � ����� � ��� �)� � � & ��% ��þ ÿ � � �� ö ÿ¿þ ú � ø ù ÿ¿þ ú � ���� ö � þ ú � ø ù �� ö � ø ù � �� � ( %�� % �� ����ç)è é ê è ë�ì í�Ťê ì)î{ï� ð��
Ñhn�g k q)l j r�� s � Ç�k l ���� ���� ��� � � % � � �� # � � ��� ��� � � ��� ��� ��� � �¤% ��� �� � % ����� � ��� � � � # � � � ��� " ��!#" � �� ��� � � % � �
, - . / 0 1 0 . 2 3 4 5 6 4 7 8 9 : , ; < 8 = - 9 7 > 6 ? @�A�B 9 3 C - 2 0 - > D E E F G $
% \ ['&S^ NcT¿\ _cP(�) *�+,) - . * / 0 1 ) +32#/ +54 6 . ) 0,7 * / 8 9 - :; ) 0 < = * 7 = 0 ; =�. ),/ 0#2,>4 -�? / - .@ A 4 . 9 4 0,B C DE�F . G54 . 1 = 8 = 0 1 -�) 0IH J. 9 =�K L M�K N�O P Q R Q S T S U V2#= 0 ; = WX 8 ) - - 4 Y 6 =�9 = F * 4 - . 4 ; ? ) * Z\[
] ^ _ ` a b c d e d f g f h i�j a k
l m noopmm qr str uvw sr xp
y z { | } ~ } { � � � � � � � � � � y � � � � z � � � � � �\��� � � � z � } z � � � � � � �
� ��������� ��¡I�#�¢�£¥¤ ¦§���©¨«ª���¬ � ¯®±°Iª�²�³��´±µ 9 / . 9 / 8 8 = 0 -�. )�¶�·�¸¥4 ?�= / ; 9,= 1 7 =�? / 4 6 - A 4 . 9�8 * ) Y / Y 4 6 4 . ¹�Hº´¼» 9 = ) * = +\W½ ?�= / ; 9,= 1 7 =�4 -�8 * ) 0 =�. )#? / 4 6 F * = A 4 . 9�8 * ) Y / Y 4 6 4 . ¹�H�:. 9 = 0
A 9 = * =¾¾ ¾¾ ¿ +5/ À 4 +,F +3< = * . = À,1 = 7 * = =�4 05ÁÂÄÃ 0 F +�Y = *�) ?�= 1 7 = -Å Ã 1 4 / +,= . = *�) ? Á
ÆÇÈÉÊË ÌÍÎÍÏ ÐÑÒÓ Ô Õ ÖÓ Ô ÕÒÒ ×××Ø�ÙÚ�Û Ü�Ý
y z { | } ~ } { � � � � � � � � � � y � � � � z � � � � � �\��� � � � z � } z � � � � � � Þ
ß ��à�¡ °Iá ��⯪ ß ������á ���������äã ��å'ª�á �æ±çéè ê ë ê ì í î ï ð ñ ê ì ò ë è�ó ì ð ô î ê õ,ö÷¥ø ò ñ ê ë ö í�ù ê ú ð û üéí è ê ë ú ù ý ì ð ô ð ú ù þ í ì ô ò ú ì í ì ò î ÿ�� ò ù ú ì ò ô � ú ê ��ò ë�í ë�� ë � ë ð ��ëï ð ë ú ò ë � ð � ù ì ê è ò ð ë��\ð û ó ð ù ò ú ò ñ ê�õ�ê í ù � ì ê÷¥ø ð í î ö � ê ú ê ì õ�ò ë ê�í�ì � î ê�ð û õ�ð ú ò ð ë�û ð ì ê í ï ��ì ð ô ð ú þù � ï ��ú � í ú í û ú ê ì û ð î î ð � ò ë è�ò ú û ð ì í�ó ê ì ò ð �� þú � ê�ì ê è ò ð ë��\ò ù�ï ð ñ ê ì ê � þ÷ ú � í ú ò ù ö û ò ë � þ û ð ì ê í ï ��ì ð ô ð ú þ í�û ê í ù ò ô î ê�ù ê ú ð û ó ð ò ë ú ù öù � ï ��ú � í ú
��� ê ì ê � � � ��ò ù ú � ê�í ì ê í�ï ð ñ ê ì ê ��û ì ð õéó ð ò ë ú � �÷�� ë�í ë ú � ì ð ô ð ú ò ï ù þ ù í õ�ê�ó ì ð ú ð ï ð î í ó ó î ò ê ù ú ð�í î î ì ð ô ð ú ùæ � ù ù � ê ù ö÷ ç î è ð ì ò ú � õ÷�� ð � ë ��ð ë�ï ð ñ ê ì ú ò õ�ê�� ò ú � üéí è ê ë ú ù þ � � ü �÷�� ð ô � ù ú ë ê ù ù ú ð�ö��� ` � � f � _�` b b c b ���� � ^ f � f ^ � e g e f g � b ` �
!"!#$
%& $(')*+ ,-, ./.00
1-12343 5 6 6 6 5743 584 9::: ;;;
y z { | } ~ } { � � � � � � � � � � y � � � � z � � � � � �\��� � � � z � } z � � � � � � <
=?>A@CBEDF G�HJIJ=?>CG�K F G�L�>AL�MON�>CPQICF G´QRTSVU W S S XEYCZ X [ \ B ] S XEU W S S X�^ W Z B Z _ Z \ W Z ` W a´cb�d aEX S `/S e�f Z BCg a B g aES ` g e Z f W a g h a _ Z ^/S X Z e \ B ] e X Z f a g h Z B ij B f S _ a X a i Z X a ZTk a l ] l i \ X e m´Qn e�f Z BCg a B g a�\ B e d aEf W a Z B \ B ] Z X a ZTZ B i ZW \ e e W a�`/a o S B i lprq s tuq v�w(xzy { | sC} pr~zx��~z� � y s � � ���J�� � � � � � � � �����A��� � � � � �C��� � � � ����z��� ¡z¢ £ ¤ ¥ ¦ § ¦E¨ © ª «T¬ « ® ¯ ¦ § ¦ °± ® ¨ « ¤ ¨ «T¤ ¥ ¦© ¦ « © ¨ « ²Tª § ¦ ª ³´/µ & ¶ · ¸ ·¹ º » ¼A½ · º ¾ ·E& ¸ º ¿ · ÀÁ/ ¤ ¥ ¦ § è © ¦ ³Ä Å ·E& ¶ ·E& ¸ º ¿ ·E& µTÆ º ¿ Ç & ¸ º ¿ ÇÁ ¢ £ « ® ÃE¥ ¦ § ¦�¤ ®EÈ ª É ¤ § ª É/³Å & µ Ê
Ë Ì Í Î Ï Ð Ï Í Ñ Ò Ó Ô Õ Ó Ö × Ø Ù Ë Ú Û × Ü Ì Ø Ö Ý Õ Þ ßCà á Ø Ò â Ì Ñ Ï Ì Ý ã ä ä å æ ç
èêéìëîíïéñðóòuô õ�öT÷ öøcùzú û ü ý þ ÿ � � �������� � ûVÿ ����������� � ü������ ÿ ��� � ÿ � � ��� ú���� � ��� ý ÿ ��! ��� û üCþ ü ý � � � ü ûVÿ ý ü�� � ü � � ý � � ü û#" � ü ��� ü ý ûø%$ü ����ÿ��& ÿ � ��' (�) *�+�) ,Cû � ü �� � þ ü ý ûVÿ ��� ü ÿ û �-+ . /0� ü ! ÿ ý ü ÿø%12� ü � ý ü ��! � ü ý ü��3547698 :�; 8 < =?> @BA�6�C D E CGF H I E D E J�K L�CGD H K H M�N O�H O E2P M E Q3SR?6�D H K H M T PUD C J N V P W X�6�Y2; :�Z [9; \ ; :�< =?>ø%]���� ü�� � � üCû � ÿ üCÿ ^ ^ ü � � ûG� � þ ü ýU� � ��ü þ�� ÿ-_
`a
bcdfe2g#h ijklnmo
Ë Ì Í Î Ï Ð Ï Í Ñ Ò Ó Ô Õ Ó Ö × Ø Ù Ë Ú Û × Ü Ì Ø Ö Ý Õ Þ ßCà á Ø Ò â Ì Ñ Ï Ì Ý ã ä ä å æ p
qsr�t�u v t�w�r�w�x5y�r�z%{�v t}|S~�r�t�u �
ø71�� ü ý � ú�����ý ü � � ý � ûG� ��� ��� � � ÿ �� ���� ��� �� �?� � �ø7��ÿ û � ü ýU� � ÿ �� ý ü ��� � � ü ��ú����������ø ù� �� � ��ÿ �� ÿ � ��� ûG]��f� ÿ ý ���ø�� ü � û � � � þ ü � ��� � ÿ ��� ü ûG� ��^ � ��� ý �� ÿ ���ø7������ ü ý ÿ � � þ üCú ü � ÿ þ�� � ý � ûG ý ü û ü ����ú����û ü ü ���� �� ûG� � ! ü ý��ø71�� ü ý ü û ��� � � ����� ý ÿ � ü � ûVÿ�� ��+������ � �� ) � �2��^U� � üCý ü ��� �����
� � �9�?� ��� � ���
� � � � � � � � ¡ ¢ £ ¤ ¢ ¥ ¦ § ¨ � © ª ¦ « � § ¥ ¬ ¤ ®�¯�° § ¡ ± � � � ¬ ² ³ ³ ´ µ µ
¶¸·º¹¼»�½2¾f¿s¹
ÀÂÁ�ÃsÄ Å Æ?Ç È�É�Ê É�Ë Ì Ë Í Î Ë Ï�Ð�È�Ñ�Ò Ç Ë Ó�Ô�Õ ÖU×�Ê Ç Ê Ó�Ø�È�ÙÛÚ�Ê Ì Ü�Ë Ó�ÝÀ%Þ2ß�à�Ò Ï Î Ò Ø�Ï È�Ñ�Ò Ç Ë Ó�Ô�Î Ë Ù�Ò Ë ÍGÉ�È�á�Ó�Ø�Ò Ø�â
ã�ä#å�æ?ç è5ésê�ëë ëë�ì í�î�ï ð?ñ ò ó�ì í�î�ï ðô�õ�ö ÷ ö2ø Ë ÍGÎ ù�Ò Ù�Ê ß�Ë Ù�á�ÙÛÇ Ò Í Ë Í Î Ê Ó�Ï Ò Ë Ó�ú�ûË üUÙ�Ê Ø�Ò�È�üUÊ Ù�Ê Î Ò Ç Ë Ê Ì#Ú�Ë Î ù�á�Ó�Ë Î ý�Í ù�Ò Ò Î þ Ç Ò Í Ë Í Î Ê Ó�Ï ÒÄ Ê Ì Ú�Ê ý�ÍGÌ Ò Í ÍGÎ ù�Ê Ó�Î ù�Ò Ø�Ë Ê Ù�Ò Î Ò Ç Ö
À%ÿ�Ò Ó�Ï Ò�âÆ?Ð0Ë Í�� � � ô�ö ÷ Î ù�Ê Ó�����Ð0É�á�Î�Ù�á�Ï ù�Ù�È�Ç Ò ÷ ��� � Î ÈGâ�� �� � ��� � � � ����� � � � ��� � � � � � !� � � � � ���"#� � � $ � � � � � ��� � � � � �
À&%�á�ÇUÒ ß�à�Ò Ç Ë Ò Ó�Ï Ò�â �'��Ð)(#Æ?Ð0ù�ý�É�Ç Ë Ø�Ë ÍGÉ�Ò Í Î
� � � � � � � � ¡ ¢ £ ¤ ¢ ¥ ¦ § ¨ � © ª ¦ « � § ¥ ¬ ¤ ®�¯�° § ¡ ± � � � ¬ ² ³ ³ ´ µ *
+',�-#.�/#-10 2 314!5�67-#8:9�5 0 8!;<7=�=�-1/)>�.!9�?�@�8A.!?B31.!C�-#/)0 2 D&-)>�EF4!?�0 G H�4�5 I�8
J KML Ê Ä Ô�Ò Ö Ó�Î Í û ø K Ç Ò Í Ë Í Î Ê Ó�Ï Ò ûON K à�Ò Ç Ë Ù�Ò Î Ò Ç ûP KML Ñ�Ò Ç Î Ë Ï Ò Í ûRQ KML Ò Ø�Ô�Ò Í û&S K Ø�Ë Ê Ù�Ò Î Ò Ç û)T K ü Ê Ë Ì á�Ç Ò à�Ç È�É�UV K Î È�Î Ê Ì#Ê Ç Ò Ê ûBW K Ç È�É�È�Î X Í�Ê Ç Ò Ê û ÷ K Ç È�É�È�Î X ÍGÇ Ê Ø�Ë á�Í
Y ZM[ \:] ^ _ ` a b[ \c] ^ _d e d fghY i] jckml n o [ \c] ^ _p q r s tumvwh
x e d p s d y y z
{#| [ } k ~ _ ] [ } � ~ _f � � f ��cv#�� | [ } k ~ _ ] [ } � ~ _� f q p s x f ��#v#�
� q r ���
� � ������ � �:� ��� � � � � � � �:�� � ��� � �
� � � � � � � � ¡ ¢ £ ¤ ¢ ¥ ¦ § ¨ � © ª ¦ « � § ¥ ¬ ¤ ®�¯�° § ¡ ± � � � ¬ ² ³ ³ ´ µ �
�����m�A���� '�m¡�½£¢ ¤¦¥§��
¨ Ò�Ò ß�Ê Ù�Ë Ó�Ò Ø�Î Ú�È�Ú�Ê ý�ÍGÎ È�Ë Ù�à�Ì Ò Ù�Ò Ó�Î�Ò Ñ�Ê à�È�Ç Ê Î Ë Ó�Ô�Î Ç Ê Ë Ì Í?âÀª©�Ë Ç Î á�Ê Ì�Î Ç Ê Ï Ò ÍGË Ó�Ç È�É�È�Î X Í�Ù�Ò Ù�È�Ç ý2â����Ð0Ç È�É�È�Î#Ú�Ë Î ù�ü Ë Ø�á�Ï Ë Ê Ì ÍÄ «�¬��Ê É�û�ÐA¬�®�Ò à�Î U û¯�Ò Ï ù�Ó�Ë È�Ó�Ö
À Þ2Ñ�Ê à�È�Ç Ê Î Ë Ó�Ô�Ì Ë °�á�Ë Ø�È�Ó�Î ù�Ò ü Ì È�È�Ç�â©�� ¨ Ç È�É�È�Î�Ú�Ë Î ù�Ò Ñ�Ê à�È�Ç Ê Î Ë Ó�Ô�Î Ç Ê Ï Ò ÍÄ � Ò Ï ù�Ê Î Ç È�Ó�Ë Ï Í��Ê É�û� Þ±®�Ò à�Î U û¯�Ò Ï ù�Ó�Ë È�Ó�Ö
� � � � � � � � ¡ ¢ £ ¤ ¢ ¥ ¦ § ¨ � © ª ¦ « � § ¥ ¬ ¤ ®�¯�° § ¡ ± � � � ¬ ² ³ ³ ´ µ ²
¶¸·º¹³���� '�m¡)´µ�·¶�� �f½#�&� �m��¸
¹ º » ¼ ½ ¾ ½ » ¿ À Á Â Ã Á Ä Å Æ Ç ¹ È É Å Ê º Æ Ä Ë Ã Ì Í!ÎÏ Æ À Ð º ¿ ½ º Ë Ñ Ò Ò Ó Ô Õ
ÖØ×�Ù7Ú ÖØÛÜ ÝÞ×�ßMàmÙ�áâã Ü ä1å�áæMáç1è ßmç éMã!ê ëmì!Ûæíïî�ðwñ�ò ó ô�ó õö ÷�ø î�ð1ùwú ô�û ü ý þ ÿí���þ ó �!ý õ ò � ú � ò ý ý�� û � � û õ � ó � � ü � � � ý � � � � � � þ � � � û � üû � þ��!ý �!ó ò � � ý þ��!û ò � � � ü
í���� õ ������ � � ��� � ! " # � $ %'&(�" � $ ! ��)*� + , - &.�� / � !*(�� 0 1 &�21 0 � '3 � 1 � � � 4
5 6 7 8 9 : 9 7 ; < = > ? = @ A B C 5 D E A F 6 B @ G ? H I�J�K B < L 6 ; 9 6 G M N N O P Q
R S R T U V W V X Y Z X [ \ ] ^ _ ` Wa bdc e fhg'i j k�ldc m�j k n o j p q�r�p s�t c u i c n v2j w2x2p e p k2yzr�{ |�l j } u i ~
� � � � � � � � � � � � � � � � � � � � � � � � � � � � � ����� � � � � � � � � � � � � � �
��� ��¡2¢ £�¤�¥ � �
¦�§�¨ª©�«¬§�¨®d¯�°²±³§�´�µ ¶·´¸�¹ º�»�¼*½ ¾ ¿�ÀÂÁÃÁ ¿2»�ĸ¾ ¿�Å�°2Æ��¯2¾ �ºdÅ�Á »dÇ�§�¯�¾�´h»�Æ�Å*¼*½ ¯2º�©�´�È�É*¹ ½ ¯�ºÊ�¯�°¶·d¾ ¼*»�¹ ¹ ½ ¯�ºdË
ÌÎÍÐÏ�Ñ�Ò Ó Ô Õ×Ö Ø Ô Ù Ú×Ï�Û×Ù�Ü×Ï�Ý Ô Ù Ô Ï�Õ×Ô Õ×ÖÌÎÞÐÛ×Ù Ï*ßáà�Ù Ô â ã*ßáÏ�ä×Û×å à�Ó�â Ï*Ï�Ü×Ò Ó à�Ù Ô Ï�ÕÌÎæ×Ô ßáÜ×å Ô â Ô Ù çÃÒ Õ×à�è×å Ò Ý�Ó Ò ä×Û'Õ×ä×à�Õ×â çÃà�Õ×äéÓ Ï�è×Û×Ý Ù Õ×Ò Ý Ý Ù Ï
Ý Ò Õ×Ý Ô Õ×Ö Ò Ó Ó Ï�Ó Ý ã�Ó Ï�è×Ï�Ù�ê à�Ô å Û×Ó Ò Ý�à�Õ×äéÒ Õ×Ñ�Ô Ó Ï�Õ×ßáÒ Õ×Ùdâ Ú×à�Õ×Ö�Ò ÝÌÎëÐÒ Ù�ìÃßáà�Õ×ç í×Û×Ò Ý Ù Ô Ï�Õ×Ý�à*Ó Ò�Ý Ù Ô å å'Ï�Ü×Ò ÕÂî
� � � � � � � � � � � � � � � � � � � � � � � � � � � � � ����� � � � � � � � � � � � � � ï
ðÐñ¬òôó³õ�öÊó·÷ùø�úÊó·û ü*ý ñé÷Êû
Ìÿþ²Û×å Ù Ô �ÐÏ�è×Ï�Ù�� ��� � ��� ��� Ü×Ü×Ò Ó·è×Ï�Û×Õ×ä×Ý�Ï�Õ��� � �2ê Ï�Ó��2Þ�� ã��ÐÞ�� à�Õ×äéþÂÞÐÍ
Ì���Ú×Ò�Ò ê ê Ò â Ù�Ï�ê�� � ����������� � � â Ú×Ï�Ï�Ý Ò�Ù Ú×Ò�Õ×Ò ��Ù�Ý Ù Ò ÜéØ Ô Ù ÚáÜ×Ó Ï�è×à�è×Ô å Ô Ù ç� Ô Õ×â Ó Ò à�Ý Ò·Ó Ï�è×Û×Ý Ù Õ×Ò Ý Ý�Ù ÏÃÕ×Ï�Ô Ý çÃÝ Ò Õ×Ý Ï�Ó Ý! �"
ÌÿÞÐÜ×Ü×å çÃÙ ÏÃßáÏ�Ó Ò·â Ï�ßáÜ×å Ò �#��� ��$�% � ��� Ò & Ö�&�à�Ý�à�Õéà�Õáà�Ü×Ü×Ó Ï�à�â ÚéÙ Ï â Ï�Ï�Ü×Ò Ó à�Ù Ô Ñ�Ò�Ü×Ó Ï�è×å Ò ß Ý Ï�å Ñ�Ô Õ×Ö' ( ) * * + , + - .�* , / 021 34- .657/ 8 / - 9;:4< . ) 0;) = > . ) / - ?!' 3457:4@ A * / B - ( + (!C 9 AD / , ) E /!+ .�> 8 A B F ) - E!B F +�E 8 / CG> 8 H - / I78 + ( E + @
Ìÿæ×Ú×à�Ü×ÒKJ ��� ��� L M ��� è×Ó Ô Õ×ÖÃÙ Ú×Ò·Ö�Ó Ï�Û×ÜéÙ Ï àÃä×Ò Ý Ô Ó Ò äáâ Ï�Õ×ê Ô Ö�Û×Ó à�Ù Ô Ï�Õ×ãØ Ô Ù Ú×Ï�Û×Ù�â Ò Õ×Ù Ó à�å*â Ï�Õ×Ù Ó Ï�å
� � � � � � � � � � � � � � � � � � � � � � � � � � � � � ����� � � � � � � � � � � � � N O
ðÐó�P ó�Q·ü'ó�RTSÂú2U2P ý Q�Véü*ý ñé÷Êû
WYX�Z [K\ Z ]�^ _ \ \ ` ^ a b c�b d d e�c d�c \ ` ^ a f Z \ ^ _ g!_ a h ]!ij k�l m;l n o p q r s t u v w x m;y t z u {|y t v q y } ~ w t u } } v � u w q u oG� � � � �W�� �4] b b ^ c�^!h d ���6� \ ^ \ a d _ ^ b ` ] c d � ` h ]|�
^!b ] c c d _!��]�h ^ _!b ] ^ ` _ � ` d ��\ Z ]|^ _ \j ~ l k�l �Ky � w u o x m;l � v w � u w � y p {|x k�l m!l n o p q r s t u v w x ~ � � �G��� � � � �W��|�4� �4���4� fG] � _ \ c � �7] \ ��d ` e c �G� ` ] ] c�^ _ g4� � � f ` ^ ��Z cj ~ l k�l �Ky � w u o x m;l � v w � u w � y p {|x k�l m!l n o p q r s t u v w x � �6�G�6� �W¢¡K�7£�¤ c ¥ ¦�£#§ ¨;] \ ] ` �4a _ a c �©^ _ g4ª6^ _ g d �7_ ] c c�^ c�£|d ���|b ] �4] _ \ ^ ` [�G� �6` d ^ h Z ] c�\ d7ª|d � d \ a h|«G¬ �6b d ` ^ \ a d _!d �!£|d _ \ a _ � d � c!6_ e _ d �|_�¨;d �7^ a _ c® ~ l k�l �Ky � w u o x m;l � v w � u w � y p {|x k�l m!l n o p q r s t u v w x�~ ¯ °G°!� �
W��#¨;a c \ ` a � � \ ] g���_ \ ��b fGd ` a \ Z �;� d `�« � � a h a ] _ \ b [!¦6^ \ ` d b b a _ fK^��4] \ ��d ` ej ±�l ²�y w ³ ´ s r v x ~ l k�l �Ky � w u o x k�l m!l nGo p q r s t u v w x kGµ�� � ¶ � �W 34- - > 8 F6/ *G·©> . ¸ + 0;> . ) ¹ F|> - (;34, . ) * ) ¹ ) > 8 º - . + 8 8 ) E + - ¹ + A» ¼ ½ ¾ ¿ À Á ¿   à ½�Ä Å ��_ \ ª|d � d \ a h c® ~ l k�l �Ky � w u o x k�l m;l nGo p q r s t u v w�Æ6ÇG� v t ³ o s x � � �
È�ÉGÊ Ë�Ì Í Í Î Ï Ì ÐGÑ6Ò�ÓGÔ Õ�Ö Ð × Ð Ø Ù Í × Ú Û Ü Ï Ý Þ Î Ø Ý Ð Ê Ì ß Þ Ð Û × Ì Í Ì Î Ê Ë�Ð Ê à Ï Ý Ê Í Ì Ý Ñ6Û Ï Î Ø Ð × Ì áâ ã ã ä6å æ æ ç�ç�ç6è é ê è ã ë é â ì í î ì è ï é è í ð æ ñ çGï ò ì ë ó æ
David PaytonRichard VaughanHRL LaboratoriesUSA
2/18/2003
1
Virtual Pheromone Robotics
[payton,vaughan]@hrl.com
David PaytonRichard Vaughan
OverviewWe seek the benefits of Ant Algorithms while using more convenient robot communications technologyWe describe two systems that perform realistic tasks in unknown indoor environments
1. An ant-like resource transportation team2. A swarm of simple robots that locates a target Both use conventional comms (serial IR and
802.11 wireless) to create environmental gradients
2/18/2003
2
System 1: Ant-like Resource transportation
Team of N robots starts from a `home’ location A in an unknown environmentExplore to find supply of resource at BCarry resource home and return for moreExemplified by foraging of ants & bees
Insect trail-following
Ants lay & follow directional chemical trails -stigmergyHoney bees communicate parameters of successful missions to their sisters Pros: adaptive, robust, distributed, scalableCons: not necessarily optimal, emergent, tricky mechanism
2/18/2003
3
Trails in localization spaceAbstract away the chemical trailCommunicate landmarks in shared localization spaceWill work as long as localization spaces are correlated to within the system’s noise tolerance Pros: adaptive, robust, distributed, scalable + lightweight, simpleCons: not necessarily optimal, robots must be localized (not trivial indoors, but many options exist)
Crumb trailsA crumb is a landmark/signpost in localization space, indicating the direction to move nextTrail represented by a linked list of “crumb” data structuresEach crumb contains:
a position (x,y)a suggested heading θa timestamp t
2/18/2003
4
Trail laying algorithm
Obstacle
Trail reading algorithm
Robot
Suggested movement = vector sum of crumbs within sense radius
2/18/2003
5
Simulation study
Trail following outperforms random walk and directed searchTolerates significant localization errorEmergent resistance to interference
Whistling in the Dark: Cooperative Trail-Following in Uncertain Localization SpaceVaughan, Stoy, Sukhatme & Matarić, Proc.Int.Conf Autonomous Agents 2000
Real world evaluationTransfer the earlier results to real 4 x Pioneer 2 robot team in a structured environmentUse trail-following with a different, specialized local navigation controllerCommunicate via wireless ethernet using UDPLocalize using odometryHow quickly will real-worldodometric drift cause system failure?
2/18/2003
6
ResultsFour robots ran for 27 mins, completing 28 round trips of 60m each (total 1.68km) before accumulated odometry errors defeated system.The 4 robots performed 3 times better than a single ideal-performing robot.
1 2 3 4
Self-correcting trailsRobots agree on a set of task-relevant events (e.g. get-resource, drop-resource)A crumb is a landmark/signpost in localization space, indicating the distance to a goal eventEach robot announces a crumb on the network every T secondsReceived crumbs stored in a linked list, representing a ‘trail’Each crumb contains:
an event name Ea position (x,y)a distance-to-event estimate da timestamp t
2/18/2003
7
Using the trail
Suggested movement = towards the crumb within r with lowest distance-to-goal
Results
“LOST: Localization-Space Trails in Robot Teams”, Vaughan, Stoy, Sukhatme & Mataric, IEEE Transactions on Robotics and Autonomous Systems18:5 2002
2/18/2003
8
Resource transport summaryDemonstrated cooperative search and transport
4 robots traveled 8.2km (5.1 miles) over 3hrs 10mins until batteries or other hardware failed
The trail laying/following methodis robust wrt. real-world odometry errorscales to N sources, M sinks, O population is independent of navigation strategy: therefore independent of robot hardwareBuilds just enough world model for stigmergyhas low computation and bandwidth requirements (16 bytes/second per robot)
T
T
System 2: Swarm finds hidden target
TT
Deployment Dispersion Detection
Pheromone Propagation
Path Visualization
2/18/2003
9
Key ThemesIR Virtual Pheromones
a message protocol that emulates pheromone diffusionrobot behaviors based on attraction & repulsion to gradients
World Embedded Computationthe structure of a distributed computation is directly entailed by the environment
World Embedded Displayrobots distributed in the environment become pixels contributingto a larger overall picture
Dispersion With Pheromone Propagation
Target
Pheromone messages propagate when
target is detected
Arrows show shortest path to target
2/18/2003
10
IR Virtual PheromonesPheromone
Source
Pheromone Relay
decrement pheromone strength
ignore packets of lower pheromonestrength
The virtual pheromone is inspired by properties of chemicalpheromones used by social insects such as ants and termites.
Facilitates emergent coordinatedactivity of thousands of
cooperating agentsyes
Sender needs not know identity ofrecipient
yes
Sender needs no acknowledgementof message receipt
yes
Information conveyed throughintensity gradients
Yes, but by using intensity-taggedmessage packets
Propagation is sustained passivelythrough the environment
Propagation is sustained by activerelaying between robots
Different chemical types neededfor different messages
Messages may be tagged withadditional data
Chemical Pheromones Virtual Pheromones
Directional IR Digital Receiversmodified to provide signal strength output
Directional IR TransmitBeacons(under each receiver)
CommunicationsPIC
Transceiver Hardware
MAPSINGLE
PROCESSOR
RESULT
MANYMOBILEPROCESSORS
RESULTS
WORLD
WORLD
World-Embedded Computation
The structure of the computation is entailed by the structure of the
environment itself.
Many complex graph-theoretic algorithms can be performed without creating an intermediate representation. The world is its own map.
The computation remains the same,
only the data is changed.
ABSTRACTION
DISPERSION
CONVENTIONAL
WORLD-EMBEDDED
2/18/2003
11
TT
World-Embedded Computationof Dijkstra's Algorithm
Dispersed Robots
T
1009998979695
94
93
92
Pheromone Wave Front
Local Gradient Points Toward
Objective
0/2
1/2
1/1
2/1
2/0
T1
T2
T1
T1
T2
T2T2
Using Multiple Pheromone GradientsOverlapping pheromone wavefronts allow each robot to determine which target is nearest.
2/18/2003
12
Multiple Gradients to Find Choke Points
TS
T
S
T
S
T
S
Opposing pheromone gradients drive robots toward median
User’s View Through Display
World-Embedded DisplayInformation about the environment can be conveyed to the user even though robots have no explicit maps of their locale.
Enemy 30m
Enemy 55m
Camera
Blinking Beacon
on Robot
See-ThroughDisplay
Labels andInformation
User’s View
Augmented-Reality System
2/18/2003
13
Augmented Reality MastTransmits 8 distinct codes in 8 directions
Codes convey a gradient arrow that retains constant orientation from any viewing angle
Augmented Reality HMDVirtual IO Display uses Pencil Cam For Tracking IR
User Sees Virtual Pheromone Vectors Overlaid Over Real World
2/18/2003
14
Commanding Robot SwarmsCommand Pheromone allows switching between different control systems
PDA
PDA
Narrow Infrared
Wide Infrared
• Commanding Individual Robots
• Commanding Swarms
Aim Beam with Laser
Command is not Propagated
Robots Propagate Command to Neighbors
Controlling DispersionThe handheld beacon triggers remote robots to halt and emit a repelling pheromone, creating a barrier to other robots
Barrier Pheromone
2/18/2003
15
Reinforcing pheromones reduce false positives in detecting an intruder.
Multi-Sensor Detection From Combined Pheromones
Motion Detector
Sound Detector
ConclusionWe introduce Virtual Pheromones:
frequent, small packets of data encoding a ‘distance’propagated via an artificial mediumimplement parallel, local, gradient-based algorithms
Our systems demonstrate how VP can create coordinated global behavior in teams/swarms of robotsThe example systems are very different
Broadcast vs. line-of-sight commsAnt algorithm vs. Dykstra’s algorithmAbstract (localization) space vs. world-embedded
The VP technique:Combines attractive features of Ant Algorithms with convenient communications technologyExtends to non-ant algorithms
Andrew RussellMonash UniversityAustralia
1. Ant robot hardware
In order for a robot to be able to work with physical chemical trails it will require asensor system to detect and accurately determine the location of chemical markings on
the ground. The robot will also need the means to deposit volatile chemicals in acontrollable way.
1.1.2 Tin oxide sensors
There are many sensors that respond to volatile chemicals. However, very few aresuitable for use in robotic applications. Taking into account physical size, response time,
sensitivity, cost, robustness, and power consumption, three technologies are commonly
used in robotics. Probably the most convenient of these chemical sensors are those basedon the changing conductivity of tin oxide.
1.1 Chemical sensors
Tin oxide sensors are available commercially. They are inexpensive, small in size, easy
to interface and the latest miniature devices made by thick-film technology have
relatively low power consumption. Some research groups have manufactured their ownsensors. The tin oxide sensor element is heated to about 300˚C where the presence of
reducing gasses causes a drop in sensor resistance. Sensitivities of the order of 1 ppmare claimed and sensors have been developed with responses optimized for solvents,
halocarbons, and a range of combustible gasses. The relationship between sensor
resistance and the concentration of detected gas is non-linear. Over a limited range ofconcentrations this relationship can be approximated by (Watson, 1984):
R≅ KRoCα (1.1)
where
α = sensitivityR = sensor resistance on exposure to target chemical
Ro = sensor resistance in clean air
K = scaling constant
C = concentration
As shown in Figure 1.1, a simple potential divider provides a signal that can be directly
digitized by an analogue to digital converter.
Figure 1.1 Circuit to energize the Figaro TGS800 air quality sensor (redrawn from
Figaro advertising material).
1.1.2 Quartz crystal microbalance
Quartz crystal microbalance sensors use a quartz crystal as a sensitive balance to weigh
odor molecules. A chemical coating on the crystal is chosen to have a specific affinityfor the target odorant molecules. When air containing molecules of this odor is drawn
over the crystal some of the molecules become temporarily attached to the coating. Thisincreases the effective crystal mass and lowers its resonant frequency. A simple model
proposed by Sauerbrey predicts the effect of small amounts of added mass (Lu, 1984):
∆f = −2 f 2∆m
ρv(1.2)
where:∆f = change in crystal resonant frequency
f = crystal resonant frequency∆m = increase in mass of the crystal per unit area,
ρ = density of crystal materialv = velocity of sound waves in the crystal material
A crystal coated with silicone OV-17 and camphor as target odorant provide a good
combination for robotics experiments (Deveza, et al. 1994). However, a wide range of
different coatings have proved useful. Here is a list of coating materials given byMoriizumi, Nakamoto, and Sakuraba (Moriizumi, et al. 1992):
• dioleyl phosphatidylserin,
• sphingomyelin (egg),• lecithin (egg),
• cholesterol,
• polyethylene glycol,
• ethyl cellulose, and• acetyl cellulose.
For the quartz crystal microbalance frequency shift is reported to be proportional to target
odorant concentration. The circuits shown in Figure 1.2 produces a digital output at thedifference frequency between a reference 10MHz oscillator (SG351P) and the sensor
oscillator (HA7210IP). The low-pass filter on the output of the SN74HC74 removes
spurious high frequency transitions caused by metastable failure.
Figure 1.2 Quartz crystal microbalance sensor circuit.
1.1.3 Conducting polymer sensors
Conductive polymers have also been used to make chemical sensors for use in robotics.A cheap and safe method of making polypyrrole has been developed using
electrochemical polymerisation of pyrrole between a pair of electrodes. The materialsand processing techniques used to make conductive polymer sensors are cheap and
readily available. Conductive polymer sensors are also easy to use for robotics
applications being very small in size and exhibiting a substantial resistance change onexposure to volatile chemicals. Typical device exposed to methanol produced a 14%
change in resistance with a response time to go from 0 to 95% of the final value of 24seconds. However, there are problems of consistency between sensors and of stability
and aging. The circuit shown in Figure 1.3 converts a change is sensor resistance into avarying frequency output. There is no d.c. component in the voltage across the polymer
sensor.
Figure 1.3 Conducting polymer sensor oscillator.
1.2 The control of airflow
Ants detect pheromone markings by tapping their antennae on the ground. At the
interface between a solid surface and a fluid there is a thin layer of fluid that is essentiallyimmobile. Volatile chemicals can only pass through this layer by a process of diffusion.
The relatively slow rate of diffusion means that the chemicals within this layer ofstagnant air do not move very far parallel to the surface and remain close to the source.
When the ant's tapping antenna penetrates the stagnation layer it can gather reliable
concentration readings for localizing the chemical markings. Larger scale robots measurechemical concentration at a much greater distance from the surface where local airflow
can affect the measurements.
The fluid flow simulation shown in Figure 1.4 demonstrates that air drawn over a sensorcrystal can originate some distance from the sensor inlet. In this case the sensor system is
very poor at localizing chemical markings on the ground.
Figure 1.4 A simulation of airflow around an aspirated sensor. Four chemical particles
originate on the ground below the sensor and four others are carried towards the sensor
by a current of air travelling left to right.
When a chemical sensor moves up-wind towards a chemical marking on the grounderroneous results are produced as the chemical is detected well before the sensor arrives
at the trail (Figure 1.6). A solution to this problem has been developed which involvesproviding an air curtain surrounding the sensor inlet. This establishes an outward current
of air that prevents external chemical getting to the sensor inlet. In Figure 1.5 a fluidflow simulation shows that the air curtain deflects chemical carried towards the sensor by
an external airflow.
Figure 1.5 Simulation of the effect of an air curtain around the sensor inlet. Chemicalparticles introduced on the left-hand side are excluded from the sensor.
The improved localization capability of the air-curtain sensor is demonstrated in Figure
1.6. Once again the sensor approaches a chemical marking in an up-wind direction. Inthis case the sensor does not respond to the chemical until it is close to the marking.
Figure 1.6 Response of a quartz crystal microbalance sensor with and without air curtainapproaching in an up-wind direction and then passing over a camphor trail at 0.2 cm/sec.
1.3 Laying chemical trails
A simple solution for depositing chemical markings on the ground is based on a felt-
tipped pen (Russell, 1999b). As shown in Figure 1.7, a plug of felt is soaked in themarking chemical (camphor dissolved in ethanol in this case) held in a reservoir. A
motorized cam lowers the pen to the floor when a trail is to be deposited.
Figure 1.7 A simple chemical applicator.
The felt-tipped pen applicator is simple but has the drawback that there is no control overthe quantity of chemical deposited. By using the metered pump arrangement shown in
Figure 1.8 there is much more control of the deposition of chemical and the supply ofchemical can be completely cut off which is not possible with the simple applicator.
Figure 1.8 Metered camphor applicator.
References
Deveza, R., Thiel, D., Russell, R.A. and Mackay-Sim, A. 'Odor sensing for robot
guidance', The International Journal of Robotics Research, Vol. 13, No. 3, June 1994,pp. 232-239.
Lu, C., 'Theory and practice of the quartz crystal microbalance', in Applications of
Piezoelectric Quartz Crystal Microbalances, (C. Lu and A.W. Czanderna eds.), Elsevier,
1984, pp. 19-61.
Moriizumi, T., Nakamoto, T. and Sakuraba, Y., 'Pattern recognition in electronic nosesby artificial neural network models', in Sensors and Sensory Systems for an Electronic
Nose, J.W. Gardner and P.N. Bartlett (eds.), Kluwer Academic Publishers, 1992, pp. 217-236.
Russell, R.A., 'Laying and sensing odor markings as a strategy for assisting mobile robotnavigation tasks', IEEE Robotics and Automation Magazine, September 1995, pp. 3-9.
Russell, R.A. Odour Sensing for Mobile Robots, World Scientific, ISBN 981023791X,
217pp. 1999b.
Watson, J., 'The tin oxide gas sensor and its applications', Sensors and Actuators, Vol. 5,1984, pp. 29-42.
2 Ant robot software
Mobile robots are being employed in increasing numbers as a means of transportingmaterial and partly finished assemblies within factories. Such robots are known as
automated guided vehicles (AGVs) and most find their way around by following somesort of permanent guide path such as a reflective tape or buried wire. Chemical trails can
serve a similar purpose by acting as a short-lived navigational marker. However, the
characteristics of the available chemical sensors means that care must be taken indeveloping the method of trail following. A number of research groups have formulated
algorithms to allow a robot to track a chemical trail.
2.1 The robot vehicle Sauro
At the Instituto Elaborazione Segnali ed Imamagini in Bari, Italy, Ettore Stella and co-
workers have built a chemical trail following robot that uses alcohol as a chemical markerand conductive polymer sensors to detect the trail (Stella, et al. 1995). This vehicle,
called Sauro, uses a pair of chemical sensors spaced 10 cm apart and located 1 cm abovethe ground to follow a 15 cm wide felt strip soaked in alcohol. The control algorithm
consists of starting the vehicle with both sensors over the felt strip. As the robot proceeds
it might deviate sufficiently that one sensor loses contact with the strip. The resulting fallin response from the sensor causes the robot to turn to bring the sensor back onto the tape
(Figure 2.1). Using this algorithm Sauro has been able to follow the tape with a trackingspeed of 60 mm/sec.
Figure 2.1 The robot Sauro follows a chemical trail using two sensors.
2.2 Robot control by neural network
Barbara Webb from the University of Nottingham investigates neural control in animals.
By implementing the control schemes using mobile robots the capabilities of the schemes
can be tested. In the area of chemical sensing Webb has developed a neuron model of antosmotropotaxis (pheromone gradient following) and tested it using a small mobile robot
built from Lego bricks (Webb, 1998). On this robot two Figaro chemical sensors playedthe part of the ant's antennae. The output from each chemical sensor was processed
through seven neurons to produce a temporal difference signal. Difference values fromleft and right sensors were then combined with a speed signal to drive the left and right
wheel motors. The result of this was to turn the robot towards higher chemical
concentrations and away from lower concentrations. Unreliable chemical readingscaused by air movements were stated to be a problem in these experiments.
2.3 Ant-like trail following
The trail-following algorithm employed by the ant Lasius fuliginousus seems to be verysimple (Hangartner 1967). The normal mode of trail following for the ant is to curve to
the left until the right antenna detects the trail and then to curve to the right until the leftantenna contacts the trail. In this way the ant maintains the trail between its two
antennae. An important feature of the ant's tracking algorithm involves turning backtowards the trail after an extended period even if the appropriate antennae does not detect
the pheromone (see Figure 2.2). This prevents the ant walking in circles when it loses the
trail. Instead the ant moves forward in the general direction of the trail sweeping widelyto left and right which gives a good chance of picking up the trail again. The following
algorithm was derived from this and implemented on a six-legged robot having twochemical sensors mounted ahead of the robot simulating antennae (Russell, 1999a).
• Take 8 paces bearing to the right until completed or until trail detected by theleft antenna.
• Take 8 paces bearing to the left until completed or until trail detected by theright antenna.
• Repeat
(If the left sensor detects the chemical trail then the sequence of 8 paces to the left is
started or restarted and stimulation of the right sensor starts or restarts the sequence of 8paces to the right)
Figure 2.2 Robot trail following based on the ant Lasius fuliginousus.
2.4 Trail following with a single sensor
If the ant Lasius fuliginousus has a damaged antennae then the algorithm described abovewill still allow it to follow a pheromone trail with its remaining good antennae. A
modification of this 'one antennae' algorithm is effective in guiding a mobile robot along
a chemical trail (Russell, et al. 1997). In order to find the trail the robot moves forwardsin a straight line ('u' in Figure 2.3). Upon detecting the trail the robot continues moving
straight forwards 'v' to cut across the trail. The slow recovery of the chemical sensormeans that edge of the trail is not detected until the robot is 'w' past the true edge. The
robot then follows the left-hand edge of the trail using the following algorithm:
• Starting away from the odor trail, rotate 'x' until the edge of the trail is detected . Once
again the slow response of the sensor means that the robot moves past the edge.• Rotate a fixed distance 'y' away from the edge of the trail.
• Move forward a fixed distance 'z' .• Repeat
This algorithm has been developed to ensure that the chemical sensor is given the maximum
time to recover after being exposed to the chemical trail. After detecting the chemical thesensor is moved clear of the trail by quickly rotating distance 'y'. The robot then waits at this
point until the sensor reading has returned close to its clean air reading before continuing.
Figure 2.3 Following a chemical trail with a single sensor.
If a robot has the ability to lay and to detect chemical markings it is possible to formulate
a number of generic applications for this technology (Russell, 1999b).
2.5 Virtual umbilical
A mobile robot may be required to explore a partially known or unknown environment. This
could be the site of an accident where no accurate maps are available or where an explosion hasdamaged the building. It would usually be essential that the robot finds its way back to the
starting point at the end of its mission. If the robot paid out an umbilical cable on the outward
journey it could follow the cable to return to its starting point. However, umbilical cables tendto snag and the equipment to pay out and retrieve an umbilical is bulky and cumbersome . The
navigational effect of the cable can be gained by laying a trail of volatile chemical as illustratedin Figure 2.4. To return to its starting point the robot simply follows the trail it has laid.
Following the trail will be made more difficult if the trail crosses itself or if the trail is crossedby the trail laid by another robot.
Figure 2.4 Finding the way back to the starting point - the virtual umbilical
2.6 Repellent marking
Many useful applications of mobile robots involve sterilizing, cleaning or searching large areas
of floor. Floor area which has already been covered can be marked with a volatile chemical.The robot will avoid the marked areas and concentrate on new areas which have not been
covered. Thus large areas can be cleaned, sterilized or searched without the need for accurate
localization of the robot. A spiral area coverage technique using a repellent marking isillustrated in Figure 2.5. To start with the robot tracks around the room keeping its left side to
the wall. When the robot meets the chemical trail it changes to keeping the trail on its left side.In this way the robot spirals in towards the center of the room covering all of the floor area. For
complicated or cluttered rooms this basic algorithm will require further development.
Figure 2.5 Efficient area coverage - the repellent marking
2.7 Pathfinder
One robot may lay a trail of volatile chemical in order to pass information to other robots. If alarge quantity of material is to be transported a pathfinder robot could establish a clear path
from the start to the goal. The path finder robot would be equipped with a range of sensors andthe 'intelligence' to find a clear path through a cluttered and changing environment. This robot
lays a trail of volatile chemical to mark its path. The bulk of the material is then transported by
simple slave robots which only have sensors to follow the chemical trail and a proximity sensorto avoid bumping into the robot in front if it slows down. In Figure 2.6 the pathfinder robot is
labeled 'A' and the slave load carrying robots 'B' and 'C'.
Figure 2.6 Marking a trail for others to follow - the pathfinder.
Is also envisaged that chemical markings could be used to warn of nearby hazards or torecruit individual robots from a circulating swarm of robots so that they can provide
assistance with some task
References
Hangartner, W. 'Spezifität und inaktivierung des spurpheromons von Lasius fuliginosus
Latr. und orientierung der arbeiterinnen im duftfeld', Zeitschrift für vergleichendePhysiologie, Vol. 57, 1967, pp. 103-136.
Russell, R.A., 'Ant trails - an example for robots to follow?' Proceedings of the IEEE
International Conference on Robotics and Automation, Detroit, Michigan, May 1999a,
pp. 2698-2703.
Russell, R.A. Odour Sensing for Mobile Robots, World Scientific, ISBN 981023791X,217pp., 1999b.
Russell, R.A., Thiel, D. and Mackay-Sim, A., 'Recruiting swarm robots using coded
odour trails', Proceeding of the International Conference on Field & service Robotics,
Canberra, 8-10 December 1997, pp. 472-476.
Stella, E., Musio, F., Vasanelli, L. and Distante, A., 'Goal-oriented mobile robotnavigation using an odour sensor', Proc. 1995 Intelligent Vehicles Symposium, pp. 147-
151.
Webb, B., ‘Robots, crickets and ants: models of neural control of chemotaxis andphonotaxis’, Neural Networks, Vol. 11, No. 7-8, October/November 1998, pp.1479-
1496.
3 Actual implementations
3.1 Implementing the Lasius fuliginousus algorithm (Russell, 1999a)
To provide an approximation to the form of an ant a six-legged robot vehicle has beenconstructed. This robot is based on Genghis, a walking robot developed at MIT (Brooks,
1989). Each of its six legs has two degrees of freedom and is actuated by radio control
servo's. Total length of the robot is 22 cm. For the implementation of the Lasiusfuliginousus algorithm two chemical sensors positioned 10 cm ahead of the robot and 13
cm apart simulate the pheromone sensing capabilities of the ant's antennae. A 68HC11microcontroller generates the walking action of the robot, monitors the two chemical
sensors and implements the Lasius fuliginousus algorithm. Figure 3.1 shows a
photograph of the robot.
Figure 3.1 The ant robot showing its two chemical sensing antennae.
One of the important advantages of the Lasius fuliginousus algorithm is the ability to
accommodate large gaps in the chemical trail. This is not surprising because thepheromone trails produced by biological ants are made up of many discrete odor patches
and probably have regular gaps. To test the capabilities of this algorithm a camphor trailwas laid on the floor consisting of a 55cm straight section, a gap of 40 cm and then a
further straight section finishing with a curve. The ant robot successfully followed thecamphor trail in both right to left and left to right directions. Figure 3.2 shows the
trajectory of the robot.
Figure 3.2 The ant robot track as it follows a chemical trail from right to left and left toright.
3.2 Following chemical trails with a single sensor (Russell, et al., 1997)
For trail following experiments using a single chemical sensor it is possible to use a robotthat is much smaller than the 6-legged ant robot. A chemical sensing robot was designed
around the 10cm diameter RAT robot. The Reactive Autonomous Testbed (RAT) robotwas designed to be sufficiently inexpensive that a new robot could be built for each
experiment. The entire robot costs less than the price of one small precision electric
motor. An air-curtain chemical sensor incorporating a circular printed circuit board fanis mounted on the front of this robot (Figure 3.3). A single 68HC11 microcontroller
controls the movement of the robot, monitors the chemical sensor and implements the'single antennae' algorithm described in Section 2.4.
Figure 3.3 The RAT robot carrying a single chemical marking sensor.
Figure 3.4 The path of RAT robot acquiring and then following a chemical trail.
The diagram shown in Figure 3.4 plots the trajectory of the RAT robot as it searches forthe chemical trail. In this figure the diamonds show successive positions of the center of
the circular robot. Upon detecting the camphor trail the robot turns and follows the left-hand edge of the trail.
References
Brooks, R.A. 'A robust layered control system for a mobile robot', IEEE Journal ofRobotics and Automation, Vol. RA-2, No.1, March 1986, pp.14-23.
Russell, R.A., 'Ant trails - an example for robots to follow?' Proceedings of the IEEE
International Conference on Robotics and Automation, Detroit, Michigan, May 1999a,pp. 2698-2703.
Russell, R.A. Odour Sensing for Mobile Robots, World Scientific, ISBN 981023791X,217pp., 1999b.
Russell, R.A., Thiel, D. and Mackay-Sim, A., 'Recruiting swarm robots using coded
odour trails', Proceeding of the International Conference on Field & service Robotics,
Canberra, 8-10 December 1997, pp. 472-476.
