Pmcsedngs
ol
he
MO2
IEEE
ln emabmal
Can e em
on
Cmbd
ApplicnDw
Sep embe 1820.2W2.G$sgow.~~d.UK
SIRIUS:
IMPROVING
THE
MANEUVERABILITY
OF
POWERED
WHEELCHAIRS
A. Ci i -Balcells,
F.
Diaz del
Rio,
C.
Jimenez,
J.L.
Se illano,
C.
Amaya,
S.
Vicen e.
Escuela
Supeno
de lngeme ia In od ica U u e sidad
de
Se llla. A da
Rem
Me cedes
dn.
41012
Se illa Spain
Fax
+34-954552759.
Email cin @alc
uses
‘Abspac -
The indoo maneu e abili y
o
powe ed
wheelchai s may he di licnl
o
bo he some
in
se e al
ci cums ances. In
his
pape , we desc ibe
an
expe imen al powe ed wheelchai named
SIRIUS,
de eloped a he Uni e si y
o
Se ille, which in oduces
some simple bu eKec i e na iga ion aids. Special
emphasis
is
placed on he implemen a ion
o
eco ded
ajec o y playback and
in
he sha ed con ol modes,
i.e. he chai ’s guiding whe e bo h he use and he
compu e collabo a e. Fu he mo e,
SIRIUS
is
an open
pla o m o essay ano he kinds
o
nne ional
o
na iga ional aids, because
i s
ha dwa e a chi ec u e
is
based
on
a comme cial
PC.
This would pe mi many
de ices ha
a e
equen ly needed by he chai d i e
o be in eg a ed smoo hly in o he chai con olle .
I.
INTRODUCTION.
Powe ed wheelchai s
a e
one he mos impo an elemen s
in he ield o ehabili a ion. Indeed powe ed wheelchai s
a e
no a luxu y o mos use s bu hei only way o li e
independen ly. Much
R&D
wo k in he ield o
ehabili a ion enginee ing is being ocused on he
imp o emen o powe wheelchai s
[l],
[Z],
[31,
[4],
[5].
In
his
con ex , au oma ion, obo ics and elema ics should
play a e y impo an ole in hc ield
o
ehabili a ion
enginee ing. Du ing he las decade ad anced wheelchai s
ha e expe ienced a spec acula imp o emen om he
echnical poin o iew, and many p o o ypes ha e been
buil
o
o e come he
limi a ions
o cu en wheelchai s,
Some o hem
a e:
di icul adap abili y,
high
cos , ou da ed
designs, imp ecise ba e y gauges, di icul handling and
lack o po abili y
[4].
Besides,
a
wide ange o o - he-shel elec onic de ices
and senso s can be ound oday a e y low cos . While, o
example, mos o hese de ices a e being inco po a ed o
au omobiles
as
luxu y
componen s, hey can play a c ucial
ole in ad anced wheelchai aids. Thus, we belie ha
esea cbe s should do a special e o
o
inco po a e hese
de ices
o
powe ed wheelchai s.
An
impo an e o
is
also
equi ed
o
b ing hese de elopmen o he ma ke
as,
o he wise,
hey
will no bene i he
inal
use s.
Du ing he las yea s in he Robo ics and Compu e o
Rehabili a ion Lab a he Uni e si y o Se ille we ha e
de eloped a sys em named
SIRIUS
ha embeds many
*
This
wnl
was
suppo ed
in
pa
by
he
Spanish
“Minis e io
de
Cimcia
y
Temologis”
h ough
p ojec
TlC-2MIO-0087-P443.
mobile obo ics ea u es
[6],
[7].
I is based on a
con en ional wbeelchai bu adding senso s and a
con olle ;
so
i can be eleope a ed o na iga e in an
au onomous way. Mo eo e
SIRIUS
can be
used
as
a
esea ch pla o m
o
es new ad anced ea u es:
GPS
na iga ion sys ems, elecommunica ion de ices, miiicial
ision-based na iga ion, e c.
In
ac ,
we
a e
wo king
nowadays
in
a esea ch p ojec in o de
o
inco po a e
elecommunica ion sys ems
o
s anda d wheelchai s, and
he possibili y o in e ac ing wi h adap i e en konmen s.
In
his
pape , and using
SIRIUS
as
a pla o m, we discuss
bow he ad anced na iga ion sys em (usually ound in
mobile obo s) can help a wheelchai use
and,
also, how
he use can aid
an
ad anced na iga ion ea u e. In sec ion
I1
we b ie ly desc ibe he main subsys ems inco po a ed o
SIRIUS
o imp o ing na igabili y. We discuss in sec ion
ID
he na iga ion aids expe imen ed un il now and inally
he conclusions a e ound
in
sec ion
N.
n.
THE
Sms
PROTOTYPE.
SIRIUS
is
a wheelchai p o o ype ha
has
been buil based
on a con en ional wheelchai , and
ha
emains open as a
pla o m o
es ing
di e en hno a ions. The p o o ype
is
in ended o be a low cos
SOM MI
domes ic chai ha
should inco po a e ad anced na iga ion ea u es.
a
mTIm
--
W
Fig.
I:
Subsys ems
o
SIRIUS
0-7803-73S3/WS17.00
0
2002
IEEE
790
Besides we can no o ge ha a pe sonal wheelchai can
ne e be a ully au onomous sys em
(like
AGV, Au oma ed
Guided Vehicles), bu a
“semiau onomous- eleope a ion”
sys em.
We mean ha we canno elimina e he use
ope a ion mainly because:
a)
complex na iga ion sys ems
a e p ohibi i e due o he in ended low cos o he
comme cial wheelchai s,
b)
ully au oma ic sys ems usually
a e e used by
use s
and
c)
In houses, he e
a e
small
innica e spaces whe e he chai mus na iga e in.
So
he
use
will
some imes ha e o guide i
(o
show he ajec oly
o
he chai he i s ime), and we a e de eloping a mixed
o
“sha ed con ol” sys em: con ol p ocessing is sha ed
be ween he
use
and he au oma ic senso -based con ol
sys em 111.
The
new ha dwa e ha he SIRIUS p o o ype includes
has
been inco po a ed in o
a
con en ional powe wheelchai .
The wheelchai
has
wo independen ly ea d i e wheels,
which cause u ning by hei di e en ial speeds, and wo
on cas o s
o
“ ee-wheels”. Below we desc ibe he main
componen s ha ha e
been
added, and in Fig.
1
we show a
schema ic o
SIRIUS.
PC-ba ed o chi ec u e.
We ha e also eached he
wheelchai p o o ype on he basis o a low cos high
pe o mance sys em.
So
we ha e “adap ed” all subsys ems
o go a ound a common CPU boa d and ba e simpli ied he
elec onic de ices needed
Ou
CPU al ema i e
is
a PC
boa d based chai . I a x86 based main boa d is used,
con ol ci cui y can be inco po a ed
as
PCI
o
ISA based
ca ds. This would also pe mi
many
de ices ha a e
kquen ly needed by he cbai d i e o be in eg a ed
smoo hly in o he chai con olle ( o example, oice
ecogni ion sys ems, mo emen
senso s,
g aphic use
in e aces, e c.). Besides, as he compu ing equi emen s
ha e inc eased, he main boa d has been easily upda ed
wi h mo e powe ul x86 p ocesso s. The only limi a ion is
ha low powe e sions should be used.
On
he o he hand,
PC a chi ec u e eases
so wa e
de elopmen , and in ac ,
he PC solu ion o embedded sys ems is being mo e and
’
mo e common
[8].
Fo PC adap a ion we ha e used
ou
p e ious expe ience
in
he
con e sion o obo con ol
uni s,
and ha e main ained
SIRIUS
as
an open pla o m o imp o e mul isenso
coo dina ion and na iga ion algo i hms.
In
his sense
he
m pu pose o
SIRIUS
was o de e mine wha na iga ion
ea u es migh inco po a e a con en ional wheelchai in
o de
o
acili a e he maneu e ing in
small
a eas.
This
will
be discussed in nex sec ion.
Powe
mo o
d i e s.
The o me chai had a pai o elays
in o de o change he di ec ion o
u n.
This
makes
impossible
o
build a high pe o mance con ol because he
swi ching delays in oduce non-linea i y when c ossing he
ze o. We inco po a ed a new powe d i e ha ensu es
linea i y in bo h di ec ions and inc eases e iciency. The
d i e is based in an H-b idge o
MOS
ansis o s, wi h a
pai o
BJT
ansis o s as he d i e s age, ha a e
commanded hy PWM wa es.
Op ical encode s
and
coun e s.
SIIUUS
ge s he posi ion
eedback h ough
wo
op ical encode s a each wheel, and
sa es his in o ma ion in
wo
16-bi s coun e s. Tha pe mi s
us
a high p ecision de e mina ion o he chai posi ion and
an accu a e eco e ing o ajec o ies in
his
esea ch
p o o ype
(as
we show
in
nex sec ion). Al hough hese
eedback senso s gi e high p ecision, we a e conside ing a
simple eedback, ha would a oid he encode s assembly
in
a
p ospec i e comme cial chai . A possibili y
is
de ec ing he in ensi y changes in he mo o s blushes,
which p ecision
will
be accu a e enough o allow he
eco e ing o ajec o ies.
Use In e lace.
Designing a comple e use in e ace needs
a
mul i-disciplina y app oach, inclndiig ehabili a ion
enginee ing, obo ics, e gonomics, psychology, compu e
science, e c.
Fo
ou
p o o ype we did no conside all hese
aspec s. We a e using he o iginal joys ick and ha e
included a 16
keys
keyboa d o selec he di e en
pa ame e s and ac ions. We ha e added a LCD display o
in o m abou he s a e o he wheelchai .
Finally,
a special
bu on ( he o me ho n) is used
o
play hack ajec o ies.
O come
all
he
inpu
in o ma ion
is
digi ized
o
be
supe ised by he CPU.
Fig.
2:
Possibili ies o low-le el eedback
i e no1 simple
senso ing
in oma ion.
Ul asound calcula ing he wo ld posi ion in a changeable en i onmen .
ansduce s
a e
he cheapes way o measu ing dis ance and Al hough
sona
in o ma ion
is
no comple e, hei dis ance
measu es can be in e es ingly exploi ed
as
explained in he
ollowing sec ion.
was
done by he
use
o
by he CPU (see Fig.
2).
When
au oma ic eedback is no ac i a ed he only adjus able
“pa ame e ” is he powe gi en o each wheel
h ough
he
joys ick o any o he use in e ace (see Fig.
2,
uppe
d awing). This makes he use he con olle , con inuously
adjus ing he joys ick depending on he ex e nal condi ions
wo king
wi h
’Ims p o ome
and
discussing
hem
o ine ia, loo slope, e c. Fo example, in he p esence o
wi h
many
and educa o s
we
ha e
O
slopes he con ol sys em au oma ically changes he powe
he na iga ion aids
o
be included in o de o imp o e he gi en
o
he wheels
in
o de
o
main ain he desi ed speed,
and o a oid he chai ’s
huns
(mo e p obable when ising a
maneu e abili y o powe ed wheelchai s.
We delibe a ely ha e a oided beacons, ixed pa hs
o
any
slope).
This “open-loop
manual
con ol” inc eases he
o he way o guiding he chai . Mos o bibliog aphy di icul y o handling he chai . In ac , wi h he
usual
open
ocuses on au onomous obo s ha ollow ixed pa hs
o
a e
loop con ol he use “closes” he eedback pa h
nI.
NAVIGATION
MODES
AND
EXPERIMENTAL
RESULTS.
guided by beacons
o
a i icial ision. The las
is
non-
iable in his momen because o i s cos and compu a ional
equi emen s.
On
he o he hand, we hink ha adjus ing
beacons
o
ixed
pa hs in a domes ic en i onmen ha may
equen ly change is no a good idea. This way o
na iga ion
only
can i in places whe e changes o he scene
can
be
es ic ed and ha ha e mo e open spaces like
hospi als
o
indus ial en i onmen s;
his
is no he case o
a pe sonal indoo wheelchai .
The
inal
p o o ype can he conside ed lie a eleope a ed
mobile obo wi h some amoun o au onomy. The
complexi y o he in e ac ion buman-compu e in he
na iga ion has implied he o ganiza ion o he unc ionali y
in
laye s o con ol. We ha e ollowed he classical laye s
p oposed in [9], bu adding he use ac ion as a new agen
ha can in e e e wi h hem. Tha is, om he mos
“in elligen ”
o
highes o he lowes we
ind
Task
planne .
Ob iously he use always mus do his
in wheelchai s.
T ojec oiy
planne .
This can be done by he use (e.g.
selec ing a ued ajec o y like a s aigh line). Bu he
compu e o example can do i i he use had selec ed
a
ask lie “play back a ajec o y”.
In
his case,
his
planne
will
e ie e he las memo ized ajec o y and
i
will
send i o he lowe le el.
T ajec oiy
gene a o .
I gene a es he nex poin
in
he
ajec o y sen by he immedia e uppe le el. Al hough
he CPU (lie in mobile obo s) usually does his
gene a ion, i he use mo es he wheelchai h ough
he joys ick, hen hdshe is cons mc ing hidhe
own
haie n
-
7
-----,
.
Con olle .
This le el ies o con e ge
eal
pos u e o
he desi ed pos u e gi en by la es le el.
F om he p eceding. o ganiza ion we can dis inguish
wo
kinds
o eedback, which
can
be al e na i ely done by he
use
o
he compu e :
1)
The low-le el eedback, done a
he con olle le el.
2)
The high-le el eedback, done a he
ajec o y planne le el
(in
SIRIUS
he
ask
le el eedback
[coope 95].
O he wise, when a compu e is
used
o conml he
wheelchai (see Fig.
2,
lowe d awing), he joys ick
command can be he linea and u ning eloci y (o he
possibili ies include con e ing joys ick commands in o
dis ance). This compu e based eedback con ol makes he
chai mo ion mo e in ui i e and independen o ex e nal
condi ions. He e he speed o he d i e wheels is sensed
in
o de
o
ob ain eedback in o ma ion. Fu he mo e, he
exis ence o an au oma ic low-le el eedback can se e o
une sys em pa ame e s o he use ’s needs. This imp o es
he adap abili y o he chai
making
possible a “ unable
con ol”. Fo u u e wo k e en he inpu signals om he
use
in e ace could also be p e-condi ioned, o example o
neu alize in en ion emo .
The second g oup o essays was based in he need
o
elie e he use . We mean o say ha
in
se e al si ua ions,
he di icul o bo he same guiding
o
he joys ick (o any
o he in e ace) can be subs i u ed o p e ixed ajec o ies.
We ha e concluded ha h ee ajec o ies we e especially
in e es ing o he
use s:
s aigh ahead o back, sligh
changes o pa h’s cu a u e, and
90
deg ees
u ns.
These
ajec o ies a e selec ed in
SIRIUS
h ough he keys
1
o
9
(‘2’
o wa d,
‘8’
backwa ds,
‘4’
90
deg ees le
u n
and
‘6’
90
deg ees igh
u n,
as
shown in Fig.
3).
When mo ing in
a co ido , he selec ion o a s aigh pa h can be e y
com o able.
In
his case i is necessa y o add some
mechanism
o
neu alize he li le o ien a ion’s de ia ions
o he wheelchai . We ha e ound
wo
simple mechanisms,
which implan a ion
will
no inc ease he wheelchai ’s cos :
sligh changes o pa h’s cu a u e ha can be chosen by he
use ( o example in keys
1,3,7
and
9
o
SIRIUS),
o
con inuous la e al dis ance measu es wi h an ul asonic
senso . Changes o la e al dis ance
will
mean changes o
o ien a ion. The las mechanism bas been ully es ed in
AGV’s,
whe e he
sona s
a e equen ly he only senso
de ice o ollow a wall
[IO].
Some imes hey inco po a e a
ull
a ay o
sona s
a ound he obo ( u he mo e hey a e
in ended o o he easons like maps building). Al hough
is always done by he use ).
In
he i s ials we show he di e ence when ac i a ing o
no he con olle eedback, i.e., when he low-le el con ol
some wo ks like
[
111 ha e conside ed his pckbili y, we
hink ha his op ion is no obliga o y in wheelchai s
because o i s complexi y and due o he p oblems
792
associa ed wi h sona s mul iple e lec ions and de ec ion
aul s (e.g. i he objec is
as
na ow
as
some able legs).
Mo eo e building a map may no be a good idea in
en i onmen s like some use s’ homes, which may
equen ly change.
(supposing he ini ial wheelchai o ien a ion is acing
he doo unde
30”,
he maximum de ec ion angle o
sona s). In addi ion he
use
ha guides he joys ick
has he imp ession
ha
“someone” is helping himhe .
OOOOl
Fig.
3:
Co espondence beh een
keys
andp ej xed
ajec o ies.
The hi d g oup o es s was done wi h senso s o
ob aining ex e nal eedback in o ma ion. Due
o
he high
p ecision o odome ic in e nal senso s (op ical encode s a
each wheel), we do no conside complica ed senso ing o
ge comple e in o ma ion, bu he simples ones, which a e
su icien
o
help he in e nal senso s’ guiding. We inally
decided
o
inco po a e ou ul asonic senso s:
wo
on al
(a le and igh sides) wi h high each dis ance and
wo
la e al. The la e al ones we e in ended o guiding
SIRIUS
h ough a co ido
as
explained be o e. The hn al ones
can be inac i e
o
p og”ed in
wo
main ways:
Obs acle de ec ion and p og essi e speed educ ion
used
o
app oach an objec acing i . We
mix
wo
di e en e ec s: eloci y educ ion and
um.
The
speed
is
educed
as
we app oach he obs acle, and each
wheel’s speed educ ion is p opo ional
o
he in e se
o he dis ance
o
he objec de ec ed by each sona .
The di e en speed educ ions ansla e in small
huns
ha aces he chai o he objec .
Obs acle a oiding wi hou changing he main
wheelchai o ien a ion ( e y use ul
o
c oss a doo )
sa ing he use om bo he some guiding. He e we also
mix
he eloci y educ ion and he
huning,
bu
in
a
di e en way. The speed is educed as we app oach he
obs acle, and a he same ime when one o he sona
de ec s a wall bu he o he does no a small
um
is
pe o med
o
a oid he doo
walls
( o example, he
au oma ic igh
u n
in Fig.
4).
The small
ums
a e
eco ded; when he obs acle a e
los
he old o ien a ion
is esumed ( he iinal le
um
in Fig.
4).
This is done
wi hou use in e en ion. Howe e i he
use
is
guiding he joys ick, he au oma ic co ec ion o
o ien a ion helps he use
o
c oss he doo . This
is
simply done by adding he joys ick command
o
he
au oma ic command (wi h a scale ac o ). The esul is
ha is almos impossible
o
collide wi h he doo walls
Le
sona
I
de ec ion
Fig.
4:
Obs acle a oiding helps o
c oss
a doo .
Howe e ul asonic senso s
su e
om se e al d awback
hey do no wo k p ope ly in e y noisy en i onmen s, hey
ha e a limi ed sensing ange wi h signi ican
minium
dis ance, e c.
Fo
hese di icul ies i
is
easy desi able ha
he sona s in o ma ion
will
be combined
wi h
o he
senso s.
The
las
g oup o na iga ion
aids
is he mos au oma ed
playing back eco ded ajec o ies. The main in en ion o
his is
o
a oid he
use
he di icul maneu e ing o e e se
d i ing, and may be e y use ul
in
small
a eas
like
ba h ooms. The e o e he CPU is always memo izing he
ajec o y done by he chai (no ma e which agen has
done i ). Fo example, in Fig.
5
he CPU will ha e
memo ized he pa h om poin
1
(ini ial posi ion)
o
2
( inal
posi ion in he ba h oom).
When
he
use
wan s
o play back he ajec o y (e.g. when
he wan s
o
go
ou
he ba h oom), he/she pushes a bu on
and he wheelchai began
o
au oma ically mo e in he
e e se di ec ion and eco e ing he
las
mo emen s. Fo
example, in Fig.
5
SIRIUS
eco e s he ajec o y om
poin
2
o
3
(posi ion a e
i s
eco e y). When he use
decides ha he pa h eco e y
was
enough (e.g. i he chai
is
ou
o he ba h oom a poin
3),
he/she can p ess one o
he s op bu ons, and he chai lea es his au oma ic mode
and s ops. F om he e he CPU con inues again
o
eco d he
ajec o y and he eco e ed pa h was
los
( agmen
2
o
3
in Fig.
5).
The e o e i in some me e s he ajec o y
eco e bu on was p essed again, he eco e y will be done
o ge ing he p e iously eco e ed pa h. Then
i
he bu on
we e p essed in poin
4
o Fig.
5,
he eco e y will pass
h ough
poin s
4,
3
and
1,
o ge ing he p e iously
793
eco e ed ajec o y (inside he ba h oom).
This
pe mi s
o
“nes ” he epe i ion o ajec o ies.
4.
Final
Posi ion
3.
Posi ion
a le
I.
Ini ial
Posi ion
i s
eco e y
2.
Final
posi ion
in
he
ba hmom
1
Fig.
5:
Playing backpa hs
in
SIRIUS:
hepossibili y
o
nes ed eco e ies
In o de
o
play again a p e iously eco ded ajec o y we
mus
choose a con ol law o he
wo
d i en wheels,
conside ing
SIRIUS
as a mobile obo . This is a classical
p oblem in non-holonomic
sys ems
lie mobile obo s [12]
and se e al solu ions ha e been p oposed. We ha e
p e e ed
o
use he “pa h ollowing” app oach which
igno es he ime
o
ollow he ajec o y [12]. In
ou
sys em
his
ime is o no conce n, and his way we a oid he
d adacks o conside ing desi ed ime e olu ion
(“ ajec o y acking” app oach) such
as
discon inui ies
in
ansien esponses. The whole p ocedu e o selec ing a
good con ol law o unicycle obo s (like
SWS)
is
explained in
[
131 and
i
is a om he scope o
his
pape .
The expe imen al esul s o pa hs eco e ies a e good
enough when hese pa hs does no ha e many o a ions.
Bu
we ha e obse ed ha he o ien a ion e o s due
o
slippage
ha e
o
be conside ed. Al hough he con ol law ensu es
ha
obo ollow he in emal memo ized pa h wi h li le
e o s, slippage e o s could accumula e and he de ia ion
om he o iginal pa h may be
4
o
5cm o each
1Om
o
pa h. No e
ha
he con olle
acks
he pa h using in emal
odome ic eedback, i.e. i can no es ima e slippage e o s.
The e o e we mus in oduce some mechanism o ex emal
eedback
o
co ec
his
p oblem. In his ield he e is
lo s
o esea ch in ully au onomous
AGV’s,
bu all
o
hem
implies a complica ed senso ing in o ma ion and
p ocessing,
as
one can imagine. Al hough he e ha e some
ials o his kind
in
ad anced wheelchai s (ul asonic a ay
in
[Ill, s e eo ision
in
[3], e c.) we ha e inco po a ed a
e y simple solu ion: he use himhe sel can in oduce
li le co ec ions
o
he au oma ic eco e y. Tha
is,
i he
use sees
ha
he chai
is
de ia ed some cen ime e s o
example
o
he igh , he/she can push he joys ick
o
he
le
o
in o m he compu e o he de ia ion. The mo e helshe
pushes he joys ick, bigge will be he co ec ion in oduced
in he ajec o y.
I
is a ac ha he use usually
o e es ima es he de ia ion and hen in a ew me e s hdshe
ac s again (in he opposi e di ec ion) un il he chai
ecupe a es app oxima ely he o iginal pa h. This human-
compu e coope a ion can be conside ed
as
ano he kind o
sha ed con ol
[I].
O iginal pa h
----_
Pa h eco e ing
Righ de ia ion:
he use pushes he
joys ick
o
he le
Fig.
6:
Sha ed
con ol
in
a
pa h eco e y.
N.
CONCLUSIONS.
he maneu e abili y o powe ed wheelchai s.
All
hese
ea u es ha e been included wi hou elimina ing he use
794

na iga ion sys ems, b) he use ’s ejec ion o ully
au onomous sys ems, and c) he di icul y small spaces
whe e he chai mus na iga e. The
main
conclusions a e
cen e ed in
a)
The ad an ages o closed loop con ol ( ia
in e nal adome ic eedback) ha educes bo he some
guiding in many cases; b) The bene i s ob ained when
mixing compu e and human con ol. He e we p esen
wo
cases: compu e co ec ion o he use guiding ia simple
sona in o ma ion ( e y e icien
o
c oss doo s) and
use
co ec ion
o
au oma ic pa h eco e y (co ec ion o
slippage de ia ions, e y di icul
o
de ec ia ex e nal
eedback).
REFERENCES
1.
Coope ,
R
A. In elligen Con ol
o
Powe Wheelchai s.
In: IEEE Eng. in Medicine and Biol. July,
1995.
2.
Bou his, G., Pino, P. Mobile Robo ics and Mobili y
Assis ance o People wi h Mo o Impai men s:
Ra ional Jus i ica ion o he VAHM P ojec . In: IEEE
T ans. on Rehabil. Engin. Vol.
4,
No.
1,
Ma ch
1996.
3.
Yode , J.D., Baumga ne , E.T.,
Sk,
S.B. Ini ial
Resul s in he De elopmen o a Guidance Sys em o a
Powe ed Wheelchai . IEEE T ansac ions on
Rehabili a ion Enginee ing., Vol.
4,
No.3
Sep embe
1996.
4.
Ci i Balcells A.,
Diu
del
Rio
F., Se illano
J.
L.,
Jimknez G.
SIRIUS:
A Low Cos High Pe o mance
Compu e ized Wheelchai . P oc. o he In . Wo kshop
on Medical Robo s, pp.
23-30.
Vienna. Oc obe
1996.
5.
ANSI/RESNA. Subcommi ee on Wheelchai s and
T anspo a ion (SOWHAT) Mee ing Minu es. Mesa,
AZ,
1995.
6.
Ci i -Balcells A., Diaz del
Rio
F., Se illano
J.
L.,
Jiminez G. SIRIUS: A Low Cos High Pe o mance
Compu e ized Wheelchai . P oc. o he In . Wo kshop
on Medical Robo s, pp.
23-30.
Vienna. Oc obe
1996.
7.
Ci i -Balcells, A., Diaz del
Rio,
F.,
Jmhez, G.,
Se illano,
J.L.
A P oposal
Fo
A Low Cos Ad anced
Wheelchai A chi ec u e. The 4 h Eu opean Con e ence
o
he Ad anc. Techn. AAATE Con e ence
1997.
ISBN
4274901831C3047
(Ohmsha).
ISBN
9051993617
(10s P ess). Oc
1997.
Thessaloniki.
G eece.
8.
Hennesy,
J.
The Fu u e o Sys ems Resea ch. IEEE
Compu e . Val
32.
Augus
1999.
9.
Cox, I.
J.
Blanche
-
An
Expe imen in Guidance and
Na iga ion o an Au onomous Robo Vehicle. IEEE
T ans. on Robo ics and Au oma ion, Vol.
7,
N0.2,
Ap il
1991.
10.
Lee,
D.
The Map-Building and Explo a ion S a egies
o
a Simple Sona -Equipped Mobile Robo . Camb idge
Uni e si y P ess.
1996.
11.
Fio e i,
S.,
Leo,
T., Longhi,
S.
A Na iga ion Sys em
o Inc easing he Au onomy and he Secu i y o
Powe ed Wheelchai s. IEEE T ans. Rehabili a ion Eng.,
Vol.
8,
No.
4,
Dec.
2000.
12.
Canndas
de
Wi , C., Khennou , H.,
Samson,
C.,
So dalen,
O.J.
Nonlinea Con ml design o Mobile
Robo s. In
Recen!s T enh in Mobile Robo s.
Ed
:
Y.F.
Zheng. Wo ld Scien i ic
1993.
13.
Dim del
Rio,
F.,
Jimenez, G., Se illano,
J.
L., Vicen e,
S.,
Ci i Balcells, A. A Pa h Following Con ml o
Unicycle Robo s. Jou nal o Robo ic Sys ems
18
7,
325-342.2001
795