Artículo	70	35	105	44	595	794	1
Científico	109	35	151	44	595	794	1
/	154	35	159	44	595	794	1
Scientific	163	35	203	44	595	794	1
Paper	207	35	232	44	595	794	1
https://doi.org/10.17163/ings.n17.2017.09	412	56	545	63	595	794	1
pISSN:	416	65	437	71	595	794	1
1390-650X	440	65	473	71	595	794	1
/	475	65	479	71	595	794	1
eISSN:	482	65	502	71	595	794	1
1390-860X	505	65	538	71	595	794	1
Mobile	140	88	214	106	595	794	1
robot	222	88	284	106	595	794	1
with	292	88	340	106	595	794	1
vision	348	88	409	106	595	794	1
based	416	88	475	106	595	794	1
navigation	112	108	223	126	595	794	1
and	231	108	269	126	595	794	1
pedestrian	277	108	390	126	595	794	1
detection	398	108	503	126	595	794	1
Robot	100	145	167	164	595	794	1
móvil	174	145	232	164	595	794	1
con	240	145	278	164	595	794	1
navegación	286	145	403	164	595	794	1
basada	411	145	483	164	595	794	1
en	491	145	515	164	595	794	1
visión	144	165	204	183	595	794	1
y	212	165	225	183	595	794	1
detección	233	165	337	183	595	794	1
de	345	165	369	183	595	794	1
peatones	377	165	471	183	595	794	1
Marco	118	196	150	206	595	794	1
A.	154	196	166	206	595	794	1
Luna	170	196	196	206	595	794	1
1	196	194	200	201	595	794	1
,	201	196	204	206	595	794	1
Julio	208	196	233	206	595	794	1
F.	237	196	248	206	595	794	1
Moya	251	196	279	206	595	794	1
1	279	194	284	201	595	794	1
,	284	196	287	206	595	794	1
Wilbert	291	196	331	206	595	794	1
G.	335	196	347	206	595	794	1
Aguilar	351	196	389	206	595	794	1
2,3,4,∗	389	194	413	201	595	794	1
,	414	196	417	206	595	794	1
Vanessa	421	196	461	206	595	794	1
Abad	465	196	493	206	595	794	1
5	493	194	497	201	595	794	1
Abstract	77	248	139	260	595	794	1
Este	316	275	335	283	595	794	1
artículo	340	275	374	283	595	794	1
propone	378	275	414	283	595	794	1
el	419	275	426	283	595	794	1
diseño	431	275	459	283	595	794	1
e	463	275	468	283	595	794	1
implementación	473	275	542	283	595	794	1
de	316	287	326	295	595	794	1
un	329	287	340	295	595	794	1
robot	344	287	367	295	595	794	1
móvil	371	287	395	295	595	794	1
con	398	287	413	295	595	794	1
navegación	417	287	465	295	595	794	1
basada	468	287	499	295	595	794	1
en	502	287	512	295	595	794	1
visión,	515	287	544	295	595	794	1
de	316	298	326	307	595	794	1
bajo	329	298	349	307	595	794	1
costo,	352	298	378	307	595	794	1
que	382	298	397	307	595	794	1
sigue	401	298	423	307	595	794	1
la	427	298	435	307	595	794	1
trayectoria	438	298	486	307	595	794	1
de	490	298	500	307	595	794	1
peatones	504	298	542	307	595	794	1
en	316	310	326	319	595	794	1
tiempo	330	310	360	319	595	794	1
real	364	310	381	319	595	794	1
usando	385	310	416	319	595	794	1
una	420	310	436	319	595	794	1
cámara	440	310	472	319	595	794	1
IP	476	310	487	319	595	794	1
a	491	310	496	319	595	794	1
bordo.	500	310	529	319	595	794	1
El	533	310	542	319	595	794	1
propósito	316	322	357	331	595	794	1
de	360	322	370	331	595	794	1
este	373	322	390	331	595	794	1
prototipo	394	322	435	331	595	794	1
es	438	322	446	331	595	794	1
la	450	322	458	331	595	794	1
navegación	461	322	509	331	595	794	1
basada	512	322	542	331	595	794	1
en	316	334	325	343	595	794	1
el	328	334	335	343	595	794	1
seguimiento	338	334	388	343	595	794	1
de	390	334	400	343	595	794	1
personas	403	334	439	343	595	794	1
conservando	442	334	494	343	595	794	1
una	496	334	512	343	595	794	1
distan-	515	334	544	343	595	794	1
cia	316	346	328	355	595	794	1
segura	331	346	358	355	595	794	1
a	362	346	367	355	595	794	1
través	370	346	395	355	595	794	1
de	399	346	408	355	595	794	1
controladores	412	346	469	355	595	794	1
PID	472	346	490	355	595	794	1
y	493	346	498	355	595	794	1
on	502	346	513	355	595	794	1
−	515	346	523	355	595	794	1
of	525	346	535	355	595	794	1
f	536	346	540	355	595	794	1
.	542	346	544	355	595	794	1
Para	316	358	336	367	595	794	1
la	339	358	347	367	595	794	1
puesta	350	358	379	367	595	794	1
en	382	358	392	367	595	794	1
marcha	395	358	427	367	595	794	1
se	431	358	439	367	595	794	1
evalúan	442	358	476	367	595	794	1
dos	479	358	493	367	595	794	1
algoritmos	497	358	542	367	595	794	1
de	316	370	326	379	595	794	1
detección	329	370	371	379	595	794	1
de	374	370	384	379	595	794	1
peatones:	388	370	429	379	595	794	1
cascada	433	370	467	379	595	794	1
de	471	370	481	379	595	794	1
clasificadores	484	370	542	379	595	794	1
HOG	316	382	339	391	595	794	1
y	343	382	348	391	595	794	1
cascada	352	382	386	391	595	794	1
de	390	382	400	391	595	794	1
clasificadores	404	382	462	391	595	794	1
LBP,	466	382	489	391	595	794	1
tanto	493	382	516	391	595	794	1
fuera	520	382	542	391	595	794	1
de	316	394	326	403	595	794	1
línea	329	394	349	403	595	794	1
como	353	394	375	403	595	794	1
a	379	394	383	403	595	794	1
bordo	387	394	412	403	595	794	1
del	415	394	428	403	595	794	1
robot.	431	394	458	403	595	794	1
Adicionalmente,	461	394	531	403	595	794	1
se	534	394	542	403	595	794	1
implantó	316	406	354	415	595	794	1
un	358	406	369	415	595	794	1
sistema	372	406	404	415	595	794	1
de	408	406	418	415	595	794	1
comunicación	421	406	480	415	595	794	1
entre	483	406	505	415	595	794	1
el	508	406	516	415	595	794	1
robot	519	406	543	415	595	794	1
y	315	418	321	427	595	794	1
una	324	418	340	427	595	794	1
estación	343	418	378	427	595	794	1
de	382	418	392	427	595	794	1
tierra.	395	418	422	427	595	794	1
Las	425	418	440	427	595	794	1
métricas	444	418	480	427	595	794	1
de	484	418	494	427	595	794	1
evaluación	497	418	542	427	595	794	1
para	316	430	335	439	595	794	1
las	338	430	350	439	595	794	1
propuestas	353	430	400	439	595	794	1
de	403	430	413	439	595	794	1
detección	417	430	457	439	595	794	1
de	461	430	471	439	595	794	1
personas	474	430	512	439	595	794	1
fueron	515	430	542	439	595	794	1
la	316	442	323	451	595	794	1
precisión	327	442	365	451	595	794	1
y	368	442	373	451	595	794	1
sensibilidad,	377	442	430	451	595	794	1
obteniendo	433	442	481	451	595	794	1
mejores	484	442	517	451	595	794	1
resul-	520	442	544	451	595	794	1
tados	316	454	339	463	595	794	1
con	342	454	357	463	595	794	1
HOG.	361	454	387	463	595	794	1
Al	390	454	400	463	595	794	1
final,	404	454	426	463	595	794	1
se	429	454	437	463	595	794	1
evaluó	441	454	468	463	595	794	1
el	472	454	479	463	595	794	1
sistema	482	454	515	463	595	794	1
de	518	454	528	463	595	794	1
co-	531	454	544	463	595	794	1
municación,	316	466	367	475	595	794	1
calculando	370	466	415	475	595	794	1
el	419	466	426	475	595	794	1
retraso	429	466	459	475	595	794	1
de	462	466	472	475	595	794	1
la	475	466	483	475	595	794	1
respuesta	486	466	526	475	595	794	1
del	530	466	542	475	595	794	1
controlador.	316	478	367	487	595	794	1
Los	370	478	385	487	595	794	1
resultados	389	478	432	487	595	794	1
mostraron	435	478	479	487	595	794	1
que	482	478	497	487	595	794	1
el	500	478	507	487	595	794	1
sistema	511	478	542	487	595	794	1
trabaja	316	490	347	499	595	794	1
adecuadamente	350	490	415	499	595	794	1
para	418	490	437	499	595	794	1
una	440	490	456	499	595	794	1
tasa	459	490	477	499	595	794	1
de	480	490	490	499	595	794	1
transmisión	493	490	542	499	595	794	1
de	316	502	326	511	595	794	1
115200	329	502	359	511	595	794	1
baudios.	362	502	398	511	595	794	1
Keywords:	77	525	128	534	595	794	1
AdaBoost,	132	525	178	534	595	794	1
HOG,	182	525	209	534	595	794	1
LBP,	212	525	235	534	595	794	1
Pedestrian	239	525	286	534	595	794	1
De-	290	525	305	534	595	794	1
tection,	77	537	110	546	595	794	1
Urban	113	537	140	546	595	794	1
Navigation.	144	537	194	546	595	794	1
Palabras	316	525	359	534	595	794	1
clave:	363	525	391	534	595	794	1
AdaBoost,	394	525	441	534	595	794	1
detección	444	525	485	534	595	794	1
de	489	525	499	534	595	794	1
peatones,	502	525	544	534	595	794	1
HOG,	316	537	341	546	595	794	1
LBP,	345	537	367	546	595	794	1
navegación	370	537	417	546	595	794	1
urbana.	421	537	454	546	595	794	1
1	71	635	74	640	595	794	1
Departamento	75	636	132	644	595	794	1
de	135	636	144	644	595	794	1
Eléctrica	148	636	183	644	595	794	1
y	186	636	191	644	595	794	1
Electrónica	194	636	239	644	595	794	1
DEEE,	242	636	271	644	595	794	1
Universidad	274	636	322	644	595	794	1
de	325	636	334	644	595	794	1
las	337	636	348	644	595	794	1
Fuerzas	351	636	381	644	595	794	1
Armadas	384	636	421	644	595	794	1
ESPE,	424	636	450	644	595	794	1
Sangolquí	453	636	492	644	595	794	1
–	495	636	500	644	595	794	1
Ecuador.	503	636	539	644	595	794	1
Departamento	75	647	132	654	595	794	1
de	135	647	145	654	595	794	1
Seguridad	148	647	188	654	595	794	1
y	191	647	196	654	595	794	1
Defensa	199	647	230	654	595	794	1
DESD,	233	647	261	654	595	794	1
Universidad	264	647	312	654	595	794	1
de	315	647	324	654	595	794	1
las	328	647	338	654	595	794	1
Fuerzas	341	647	372	654	595	794	1
Armadas	375	647	411	654	595	794	1
ESPE,	414	647	441	654	595	794	1
Sangolquí	444	647	483	654	595	794	1
–	486	647	491	654	595	794	1
Ecuador.	494	647	530	654	595	794	1
3	71	655	74	660	595	794	1
Centro	75	657	103	665	595	794	1
de	106	657	115	665	595	794	1
Investigación	119	657	172	665	595	794	1
Científica	175	657	215	665	595	794	1
y	218	657	223	665	595	794	1
Tecnológica	226	657	274	665	595	794	1
del	277	657	289	665	595	794	1
Ejército	292	657	325	665	595	794	1
CICTE,	328	657	361	665	595	794	1
Universidad	364	657	413	665	595	794	1
de	416	657	426	665	595	794	1
las	429	657	440	665	595	794	1
Fuerzas	443	657	474	665	595	794	1
Armadas	477	657	514	665	595	794	1
ESPE,	518	657	545	665	595	794	1
Sangolquí	75	667	114	675	595	794	1
–	118	667	122	675	595	794	1
Ecuador.	125	667	161	675	595	794	1
4,∗	71	675	80	681	595	794	1
Grup	81	677	102	685	595	794	1
de	110	677	119	685	595	794	1
Recerca	126	677	158	685	595	794	1
en	166	677	175	685	595	794	1
Enginyeria	182	677	226	685	595	794	1
del	233	677	245	685	595	794	1
Coneixement	253	677	306	685	595	794	1
GREC,	313	677	343	685	595	794	1
Universitat	351	677	396	685	595	794	1
Politècnica	404	677	448	685	595	794	1
de	456	677	465	685	595	794	1
Catalunya	472	677	514	685	595	794	1
UPC,	522	677	545	685	595	794	1
Barcelona	81	687	121	695	595	794	1
–	124	687	129	695	595	794	1
España.	132	687	164	695	595	794	1
Autor	167	687	190	695	595	794	1
para	193	687	211	695	595	794	1
correspondencia	214	687	279	695	595	794	1
)	282	685	290	698	595	794	1
:	290	687	293	695	595	794	1
wgaguilar@espe.edu.ec	296	687	387	695	595	794	1
5	71	696	75	701	595	794	1
Departament	75	697	128	705	595	794	1
de	131	697	140	705	595	794	1
Genètica,	143	697	181	705	595	794	1
Universitat	184	697	229	705	595	794	1
de	232	697	241	705	595	794	1
Barcelona	245	697	284	705	595	794	1
UB,	287	697	303	705	595	794	1
Barcelona	307	697	346	705	595	794	1
–	349	697	354	705	595	794	1
España.	357	697	389	705	595	794	1
2	71	645	74	650	595	794	1
Recibido:	71	711	108	719	595	794	1
01-11-2016	111	711	154	719	595	794	1
,	154	711	157	720	595	794	1
aprobado	161	711	198	719	595	794	1
tras	201	711	216	719	595	794	1
revisión:	220	711	253	719	595	794	1
27-12-2016	256	711	299	719	595	794	1
Forma	71	725	97	733	595	794	1
sugerida	101	725	135	733	595	794	1
de	138	725	148	733	595	794	1
citación:	152	725	186	733	595	794	1
Luna,	190	725	213	733	595	794	1
M.;	217	725	231	733	595	794	1
Moya,	235	725	260	733	595	794	1
J.;	263	725	273	733	595	794	1
Aguilar,	277	725	310	733	595	794	1
W.;	314	725	329	733	595	794	1
Abad,	332	725	357	733	595	794	1
V.	361	725	370	733	595	794	1
(2017).	376	725	404	733	595	794	1
«Mobile	408	725	441	733	595	794	1
robot	445	725	467	733	595	794	1
with	471	725	489	733	595	794	1
vision	493	725	517	733	595	794	1
based	520	725	543	733	595	794	1
navigation	71	735	113	743	595	794	1
and	116	735	131	743	595	794	1
pedestrian	134	735	176	743	595	794	1
detection	179	735	216	743	595	794	1
».	219	735	226	743	595	794	1
Ingenius.	229	736	270	744	595	794	1
N.	273	735	282	743	595	794	1
◦	282	733	286	738	595	794	1
17,	287	735	298	743	595	794	1
(Enero-Junio).	301	735	360	743	595	794	1
pp.	363	735	376	743	595	794	1
67-72.	379	735	403	743	595	794	1
ISSN:	406	735	429	743	595	794	1
1390-650X.	432	735	477	743	595	794	1
67	302	760	313	771	595	794	1
68	51	29	63	39	595	794	2
1.	51	66	61	76	595	794	2
Introduction	68	66	142	76	595	794	2
Autonomous	51	85	104	93	595	794	2
navigation	107	85	152	93	595	794	2
is	155	85	161	93	595	794	2
a	164	85	169	93	595	794	2
research	172	85	206	93	595	794	2
line	209	85	225	93	595	794	2
with	227	85	246	93	595	794	2
growing	249	85	283	93	595	794	2
interest	51	97	84	105	595	794	2
from	89	97	110	105	595	794	2
DARPA	115	97	151	105	595	794	2
Grand	156	97	184	105	595	794	2
Challenge	189	97	233	105	595	794	2
2004.	237	97	261	105	595	794	2
Sev-	265	97	284	105	595	794	2
eral	51	108	67	117	595	794	2
vision	71	108	97	117	595	794	2
based	100	108	125	117	595	794	2
vehicles	129	108	163	117	595	794	2
have	166	108	186	117	595	794	2
been	190	108	210	117	595	794	2
developed	214	108	258	117	595	794	2
[1–3].	261	108	285	117	595	794	2
There	51	120	76	129	595	794	2
are	80	120	93	129	595	794	2
a	96	120	101	129	595	794	2
lot	105	120	116	129	595	794	2
of	120	120	128	129	595	794	2
challenges	131	120	175	129	595	794	2
in	178	120	186	129	595	794	2
urban	190	120	215	129	595	794	2
navigation	219	120	264	129	595	794	2
like	268	120	283	129	595	794	2
perception	51	132	97	141	595	794	2
[4,	100	132	111	141	595	794	2
5],	112	132	123	141	595	794	2
obstacles	126	132	165	141	595	794	2
avoidance	168	132	211	141	595	794	2
[6,	214	132	225	141	595	794	2
7],	226	132	236	141	595	794	2
people	240	132	268	141	595	794	2
de-	271	132	284	141	595	794	2
tection	51	144	81	153	595	794	2
[8–11]	84	144	110	153	595	794	2
and	113	144	129	153	595	794	2
video	132	144	155	153	595	794	2
stabilization	159	144	212	153	595	794	2
[12–14].	215	144	248	153	595	794	2
Multiple	66	156	102	165	595	794	2
pedestrian	105	156	149	165	595	794	2
detection	152	156	191	165	595	794	2
algorithms	193	156	238	165	595	794	2
have	241	156	260	165	595	794	2
been	263	156	283	165	595	794	2
developed	51	168	95	177	595	794	2
due	100	168	116	177	595	794	2
their	121	168	142	177	595	794	2
different	148	168	185	177	595	794	2
applications	190	168	243	177	595	794	2
that	249	168	267	177	595	794	2
in-	273	168	284	177	595	794	2
clude:	51	180	77	189	595	794	2
surveillance	80	180	132	189	595	794	2
[15–17],	135	180	169	189	595	794	2
driver	172	180	198	189	595	794	2
assistance	201	180	245	189	595	794	2
systems	248	180	283	189	595	794	2
[11,	51	192	67	201	595	794	2
18,	69	192	82	201	595	794	2
19],	84	192	100	201	595	794	2
robotics	104	192	140	201	595	794	2
[20,	144	192	160	201	595	794	2
21],	162	192	178	201	595	794	2
and	183	192	199	201	595	794	2
others.	203	192	234	201	595	794	2
According	238	192	283	201	595	794	2
to	51	204	60	213	595	794	2
[22]	63	204	78	213	595	794	2
most	81	204	101	213	595	794	2
approaches	104	204	151	213	595	794	2
use	154	204	168	213	595	794	2
detectors	171	204	209	213	595	794	2
based	212	204	236	213	595	794	2
on	239	204	249	213	595	794	2
HAAR-	252	204	284	213	595	794	2
like	51	216	66	225	595	794	2
features	70	216	105	225	595	794	2
[23],	109	216	128	225	595	794	2
HOG	132	216	155	225	595	794	2
[24,	159	216	175	225	595	794	2
25],	176	216	192	225	595	794	2
LBP	196	216	217	225	595	794	2
[26]	221	216	236	225	595	794	2
and	240	216	257	225	595	794	2
some	261	216	283	225	595	794	2
combinations	51	228	107	237	595	794	2
like:	110	228	128	237	595	794	2
HOG-HAAR	131	228	185	237	595	794	2
[27]	188	228	204	237	595	794	2
or	207	228	215	237	595	794	2
HOG-LBP	218	228	264	237	595	794	2
[28].	267	228	285	237	595	794	2
In	51	240	60	249	595	794	2
[29],	63	240	81	249	595	794	2
HAAR-like	84	240	131	249	595	794	2
features	134	240	168	249	595	794	2
have	171	240	190	249	595	794	2
a	193	240	198	249	595	794	2
low	201	240	216	249	595	794	2
performance	219	240	271	249	595	794	2
in	275	240	283	249	595	794	2
pedestrian	51	252	96	261	595	794	2
detection	100	252	139	261	595	794	2
task.	143	252	164	261	595	794	2
In	66	264	75	273	595	794	2
this	80	264	96	273	595	794	2
work,	101	264	125	273	595	794	2
we	130	264	141	273	595	794	2
experimentally	146	264	211	273	595	794	2
test	216	264	232	273	595	794	2
Histogram	237	264	283	273	595	794	2
of	51	276	59	285	595	794	2
Oriented	63	276	101	285	595	794	2
Gradients	105	276	148	285	595	794	2
(HOG)	152	276	183	285	595	794	2
[24]	186	276	202	285	595	794	2
and	206	276	222	285	595	794	2
Local	225	276	250	285	595	794	2
Binary	253	276	283	285	595	794	2
Patterns	51	288	89	297	595	794	2
(LBP)	94	288	122	297	595	794	2
[26],	127	288	146	297	595	794	2
feature	150	288	181	297	595	794	2
extraction	186	288	231	297	595	794	2
algorithms	236	288	283	297	595	794	2
for	51	300	63	309	595	794	2
pedestrian	66	300	111	309	595	794	2
detection	115	300	154	309	595	794	2
with	158	300	177	309	595	794	2
Adaptive	180	300	220	309	595	794	2
Boosting	223	300	261	309	595	794	2
(Ad-	265	300	284	309	595	794	2
aBoost),	51	312	88	321	595	794	2
a	92	312	97	321	595	794	2
supervised	100	312	147	321	595	794	2
learning	150	312	186	321	595	794	2
algorithm,	189	312	235	321	595	794	2
applied	239	312	271	321	595	794	2
in	274	312	283	321	595	794	2
mobile	51	324	79	333	595	794	2
robot	83	324	106	333	595	794	2
navigation.	109	324	157	333	595	794	2
The	160	324	177	333	595	794	2
video	180	324	203	333	595	794	2
is	206	324	213	333	595	794	2
transmitted	216	324	267	333	595	794	2
via	270	324	283	333	595	794	2
Wi-Fi	51	336	77	345	595	794	2
to	81	336	90	345	595	794	2
a	94	336	99	345	595	794	2
ground	103	336	134	345	595	794	2
station	138	336	168	345	595	794	2
to	172	336	181	345	595	794	2
determine	185	336	230	345	595	794	2
the	234	336	248	345	595	794	2
motion	252	336	283	345	595	794	2
direction	51	348	90	356	595	794	2
of	95	348	103	356	595	794	2
the	107	348	122	356	595	794	2
robot	126	348	150	356	595	794	2
using	155	348	178	356	595	794	2
image	183	348	209	356	595	794	2
processing,	213	348	262	356	595	794	2
and	266	348	283	356	595	794	2
the	51	360	65	368	595	794	2
control	69	360	100	368	595	794	2
actions	105	360	136	368	595	794	2
are	140	360	154	368	595	794	2
sent	158	360	176	368	595	794	2
by	180	360	191	368	595	794	2
Bluetooth	195	360	239	368	595	794	2
from	244	360	264	368	595	794	2
the	269	360	283	368	595	794	2
computer	51	372	92	380	595	794	2
to	95	372	104	380	595	794	2
the	108	372	121	380	595	794	2
robot.	125	372	151	380	595	794	2
The	66	384	83	392	595	794	2
rest	88	384	105	392	595	794	2
of	109	384	117	392	595	794	2
the	122	384	136	392	595	794	2
article	141	384	168	392	595	794	2
is	173	384	180	392	595	794	2
distributed	184	384	233	392	595	794	2
as	238	384	247	392	595	794	2
follows:	251	384	284	392	595	794	2
In	51	396	60	404	595	794	2
the	65	396	79	404	595	794	2
next	84	396	103	404	595	794	2
section,	108	396	141	404	595	794	2
we	146	396	158	404	595	794	2
do	162	396	173	404	595	794	2
a	178	396	183	404	595	794	2
review	187	396	216	404	595	794	2
of	221	396	229	404	595	794	2
the	233	396	248	404	595	794	2
related	252	396	283	404	595	794	2
work.	51	408	75	416	595	794	2
In	78	408	88	416	595	794	2
the	91	408	105	416	595	794	2
section	109	408	139	416	595	794	2
III,	143	408	157	416	595	794	2
we	160	408	172	416	595	794	2
make	175	408	198	416	595	794	2
a	202	408	207	416	595	794	2
review	210	408	239	416	595	794	2
of	242	408	250	416	595	794	2
two	254	408	270	416	595	794	2
al-	273	408	284	416	595	794	2
gorithms	51	419	90	428	595	794	2
used	94	419	113	428	595	794	2
for	117	419	129	428	595	794	2
pedestrian	133	419	179	428	595	794	2
detection.	182	419	226	428	595	794	2
We	229	419	243	428	595	794	2
describe	247	419	283	428	595	794	2
our	51	431	65	440	595	794	2
approach	69	431	108	440	595	794	2
for	111	431	123	440	595	794	2
robot	126	431	149	440	595	794	2
development	153	431	206	440	595	794	2
in	210	431	218	440	595	794	2
the	221	431	235	440	595	794	2
section	238	431	268	440	595	794	2
IV.	271	431	285	440	595	794	2
Finally	51	443	82	452	595	794	2
the	86	443	100	452	595	794	2
results	104	443	132	452	595	794	2
obtained	136	443	174	452	595	794	2
and	178	443	194	452	595	794	2
the	198	443	212	452	595	794	2
conclusions	216	443	266	452	595	794	2
are	269	443	283	452	595	794	2
presented	51	455	92	464	595	794	2
in	96	455	104	464	595	794	2
the	107	455	121	464	595	794	2
sections	124	455	158	464	595	794	2
V	162	455	169	464	595	794	2
and	172	455	188	464	595	794	2
VI.	192	455	206	464	595	794	2
2.	51	479	61	490	595	794	2
Related	68	479	113	490	595	794	2
Works	118	479	155	490	595	794	2
In	51	498	60	507	595	794	2
the	64	498	78	507	595	794	2
literature,	82	498	126	507	595	794	2
there	130	498	153	507	595	794	2
are	156	498	170	507	595	794	2
several	174	498	204	507	595	794	2
applications	207	498	261	507	595	794	2
that	264	498	283	507	595	794	2
integrate	51	510	89	519	595	794	2
people	91	510	119	519	595	794	2
tracking	121	510	156	519	595	794	2
with	159	510	178	519	595	794	2
mobile	180	510	208	519	595	794	2
robots.	211	510	240	519	595	794	2
A	243	510	250	519	595	794	2
ground	253	510	283	519	595	794	2
autonomous	51	522	103	531	595	794	2
vehicle	106	522	135	531	595	794	2
designed	138	522	174	531	595	794	2
to	178	522	186	531	595	794	2
track	190	522	211	531	595	794	2
people	215	522	242	531	595	794	2
based	245	522	269	531	595	794	2
on	272	522	283	531	595	794	2
a	51	534	56	543	595	794	2
vision	59	534	84	543	595	794	2
was	88	534	103	543	595	794	2
proposed	107	534	146	543	595	794	2
in	149	534	157	543	595	794	2
[30],	161	534	179	543	595	794	2
in	182	534	190	543	595	794	2
this	194	534	210	543	595	794	2
work	213	534	234	543	595	794	2
the	237	534	251	543	595	794	2
person	254	534	283	543	595	794	2
to	51	546	60	555	595	794	2
be	63	546	73	555	595	794	2
tracked	77	546	109	555	595	794	2
should	112	546	140	555	595	794	2
wear	144	546	164	555	595	794	2
a	167	546	172	555	595	794	2
discriminable	175	546	233	555	595	794	2
rectangle.	236	546	278	555	595	794	2
In	66	558	75	567	595	794	2
[20],	78	558	96	567	595	794	2
a	99	558	104	567	595	794	2
mobile	108	558	136	567	595	794	2
platform	139	558	176	567	595	794	2
that	179	558	197	567	595	794	2
uses	200	558	218	567	595	794	2
a	221	558	226	567	595	794	2
multiple	229	558	264	567	595	794	2
sen-	268	558	284	567	595	794	2
sor	51	570	64	579	595	794	2
fusion	66	570	91	579	595	794	2
approach	93	570	132	579	595	794	2
and	135	570	150	579	595	794	2
combines	153	570	191	579	595	794	2
three	194	570	215	579	595	794	2
kinds	218	570	240	579	595	794	2
of	243	570	250	579	595	794	2
sensors	253	570	283	579	595	794	2
in	51	582	59	591	595	794	2
order	63	582	86	591	595	794	2
to	89	582	98	591	595	794	2
detect	102	582	128	591	595	794	2
people	132	582	160	591	595	794	2
using	164	582	187	591	595	794	2
RGB-D	190	582	223	591	595	794	2
vision,	227	582	255	591	595	794	2
lasers	258	582	283	591	595	794	2
and	51	594	67	603	595	794	2
a	71	594	76	603	595	794	2
thermal	79	594	113	603	595	794	2
sensor	117	594	144	603	595	794	2
is	147	594	154	603	595	794	2
presented.	157	594	202	603	595	794	2
The	206	594	223	603	595	794	2
mobile	226	594	256	603	595	794	2
robot	259	594	283	603	595	794	2
in	51	606	59	615	595	794	2
[21]	62	606	78	615	595	794	2
uses	81	606	98	615	595	794	2
an	101	606	112	615	595	794	2
omnidirectional	115	606	181	615	595	794	2
camera	184	606	215	615	595	794	2
and	218	606	233	615	595	794	2
laser	237	606	256	615	595	794	2
range	259	606	283	615	595	794	2
finder	51	618	76	626	595	794	2
to	79	618	88	626	595	794	2
detect	91	618	117	626	595	794	2
and	120	618	136	626	595	794	2
track	139	618	161	626	595	794	2
people.	164	618	194	626	595	794	2
In	197	618	206	626	595	794	2
[31]	209	618	224	626	595	794	2
is	227	618	234	626	595	794	2
introduced	237	618	283	626	595	794	2
a	51	630	56	638	595	794	2
robot	59	630	83	638	595	794	2
to	87	630	96	638	595	794	2
assist	99	630	123	638	595	794	2
elderly	126	630	156	638	595	794	2
people	159	630	188	638	595	794	2
with	191	630	211	638	595	794	2
a	214	630	219	638	595	794	2
Kinect	223	630	252	638	595	794	2
device	255	630	283	638	595	794	2
and	51	642	67	650	595	794	2
ROS	70	642	91	650	595	794	2
packages.	94	642	135	650	595	794	2
The	139	642	156	650	595	794	2
detection	159	642	199	650	595	794	2
and	202	642	218	650	595	794	2
localization	222	642	271	650	595	794	2
of	275	642	283	650	595	794	2
people	51	654	79	662	595	794	2
is	82	654	89	662	595	794	2
an	92	654	103	662	595	794	2
important	106	654	150	662	595	794	2
aspect	154	654	181	662	595	794	2
in	185	654	193	662	595	794	2
robotic	196	654	227	662	595	794	2
applications	230	654	283	662	595	794	2
for	51	665	63	674	595	794	2
interaction.	66	665	116	674	595	794	2
The	119	665	136	674	595	794	2
work	139	665	160	674	595	794	2
presented	164	665	205	674	595	794	2
in	208	665	216	674	595	794	2
[32]	220	665	235	674	595	794	2
deals	239	665	260	674	595	794	2
with	263	665	283	674	595	794	2
the	51	677	65	686	595	794	2
task	68	677	86	686	595	794	2
of	89	677	97	686	595	794	2
searching	100	677	139	686	595	794	2
for	143	677	154	686	595	794	2
people	158	677	185	686	595	794	2
in	188	677	197	686	595	794	2
home	200	677	223	686	595	794	2
environments	226	677	283	686	595	794	2
with	51	689	70	698	595	794	2
a	74	689	79	698	595	794	2
mobile	82	689	111	698	595	794	2
robot.	114	689	141	698	595	794	2
The	144	689	162	698	595	794	2
method	165	689	198	698	595	794	2
uses	202	689	220	698	595	794	2
color	223	689	244	698	595	794	2
and	247	689	264	698	595	794	2
gra-	267	689	284	698	595	794	2
dient	51	701	73	710	595	794	2
models	77	701	108	710	595	794	2
of	112	701	120	710	595	794	2
the	124	701	138	710	595	794	2
environment	142	701	197	710	595	794	2
and	201	701	218	710	595	794	2
a	222	701	227	710	595	794	2
color	231	701	252	710	595	794	2
model	256	701	283	710	595	794	2
of	51	713	59	722	595	794	2
the	63	713	77	722	595	794	2
user.	81	713	102	722	595	794	2
Evaluation	105	713	153	722	595	794	2
is	157	713	163	722	595	794	2
done	167	713	188	722	595	794	2
on	192	713	202	722	595	794	2
real-world	206	713	250	722	595	794	2
experi-	254	713	284	722	595	794	2
ments	51	725	76	734	595	794	2
with	79	725	98	734	595	794	2
the	101	725	114	734	595	794	2
robot	117	725	140	734	595	794	2
searching	143	725	182	734	595	794	2
for	185	725	197	734	595	794	2
the	200	725	213	734	595	794	2
user	216	725	233	734	595	794	2
at	236	725	245	734	595	794	2
different	247	725	283	734	595	794	2
places.	51	737	80	746	595	794	2
INGENIUS	337	30	396	39	595	794	2
N.	399	30	409	39	595	794	2
◦	409	28	413	35	595	794	2
17,	417	30	430	39	595	794	2
Enero-Junio	433	30	486	39	595	794	2
de	491	30	501	39	595	794	2
2017	504	30	524	39	595	794	2
We	308	67	321	76	595	794	2
propose	324	67	356	76	595	794	2
a	359	67	364	76	595	794	2
simple	366	67	393	76	595	794	2
structure	396	67	434	76	595	794	2
robot	437	67	460	76	595	794	2
(equipped	462	67	504	76	595	794	2
only	507	67	525	76	595	794	2
with	292	79	312	88	595	794	2
a	315	79	320	88	595	794	2
smartphone	323	79	375	88	595	794	2
camera	378	79	409	88	595	794	2
for	412	79	424	88	595	794	2
navigation)	428	79	477	88	595	794	2
capable	480	79	513	88	595	794	2
of	516	79	524	88	595	794	2
detect	293	91	320	100	595	794	2
and	324	91	340	100	595	794	2
track	344	91	367	100	595	794	2
people	371	91	399	100	595	794	2
orientation	404	91	452	100	595	794	2
and	456	91	472	100	595	794	2
translation	476	91	524	100	595	794	2
based	293	103	317	112	595	794	2
on	320	103	331	112	595	794	2
pedestrian	334	103	379	112	595	794	2
detection	382	103	422	112	595	794	2
algorithms.	425	103	474	112	595	794	2
In	477	103	486	112	595	794	2
contrast	490	103	525	112	595	794	2
with	292	115	311	123	595	794	2
some	315	115	336	123	595	794	2
words	339	115	364	123	595	794	2
presented	367	115	407	123	595	794	2
in	411	115	419	123	595	794	2
the	422	115	435	123	595	794	2
literature,	439	115	481	123	595	794	2
our	484	115	498	123	595	794	2
robot	502	115	525	123	595	794	2
performs	293	127	332	135	595	794	2
the	336	127	350	135	595	794	2
navigation	355	127	401	135	595	794	2
task	405	127	424	135	595	794	2
using	428	127	451	135	595	794	2
a	456	127	461	135	595	794	2
PI	465	127	476	135	595	794	2
control	480	127	511	135	595	794	2
to	515	127	524	135	595	794	2
achieve	293	139	324	147	595	794	2
relatively	327	139	367	147	595	794	2
fast	371	139	387	147	595	794	2
speed,	390	139	417	147	595	794	2
avoiding	420	139	456	147	595	794	2
braking	460	139	493	147	595	794	2
issues.	496	139	523	147	595	794	2
3.	293	166	303	176	595	794	2
Pedestrian	310	166	372	176	595	794	2
Detection	377	166	434	176	595	794	2
Algorithms	438	166	504	176	595	794	2
3.1.	293	191	311	199	595	794	2
Histogram	316	191	369	199	595	794	2
of	372	191	382	199	595	794	2
Oriented	386	191	429	199	595	794	2
Gradients	433	191	482	199	595	794	2
(HOG)	486	191	522	199	595	794	2
This	292	209	311	218	595	794	2
algorithm	315	209	356	218	595	794	2
is	359	209	366	218	595	794	2
a	369	209	374	218	595	794	2
feature	377	209	407	218	595	794	2
descriptor	410	209	453	218	595	794	2
for	456	209	468	218	595	794	2
object	471	209	497	218	595	794	2
detec-	501	209	526	218	595	794	2
tion	293	221	310	230	595	794	2
focused	314	221	346	230	595	794	2
on	350	221	360	230	595	794	2
pedestrian	364	221	410	230	595	794	2
detection	413	221	454	230	595	794	2
and	457	221	474	230	595	794	2
introduced	477	221	524	230	595	794	2
in	293	233	301	242	595	794	2
[24].	305	233	323	242	595	794	2
The	327	233	345	242	595	794	2
image	348	233	374	242	595	794	2
window	378	233	412	242	595	794	2
is	415	233	422	242	595	794	2
separated	426	233	468	242	595	794	2
into	472	233	489	242	595	794	2
smaller	493	233	525	242	595	794	2
parts	293	245	315	254	595	794	2
called	318	245	343	254	595	794	2
cells.	346	245	367	254	595	794	2
For	370	245	384	254	595	794	2
each	388	245	406	254	595	794	2
cell,	410	245	427	254	595	794	2
we	430	245	441	254	595	794	2
accumulate	445	245	493	254	595	794	2
a	496	245	501	254	595	794	2
local	504	245	524	254	595	794	2
1-D	292	257	308	266	595	794	2
histogram	311	257	354	266	595	794	2
of	357	257	365	266	595	794	2
gradient	368	257	403	266	595	794	2
orientations	407	257	457	266	595	794	2
of	460	257	468	266	595	794	2
the	472	257	485	266	595	794	2
pixels	489	257	513	266	595	794	2
in	516	257	524	266	595	794	2
the	293	269	306	278	595	794	2
cell.	309	269	326	278	595	794	2
The	329	269	346	278	595	794	2
gradient	349	269	384	278	595	794	2
orientation	387	269	433	278	595	794	2
is	436	269	443	278	595	794	2
given	446	269	468	278	595	794	2
by	471	269	481	278	595	794	2
equations	484	269	524	278	595	794	2
1	292	281	297	290	595	794	2
and	300	281	317	290	595	794	2
1:	320	281	328	290	595	794	2
I	332	313	337	322	595	794	2
X	337	316	343	323	595	794	2
=	347	313	355	322	595	794	2
I	357	313	362	322	595	794	2
×	365	312	373	322	595	794	2
D	375	313	383	322	595	794	2
X	383	316	390	323	595	794	2
D	330	339	338	348	595	794	2
X	338	342	344	349	595	794	2
=	348	339	356	348	595	794	2
[1,	359	339	369	348	595	794	2
0,	371	339	379	348	595	794	2
1]	380	339	388	348	595	794	2
I	410	313	415	322	595	794	2
Y	415	316	420	323	595	794	2
=	425	313	432	322	595	794	2
I	435	313	439	322	595	794	2
×	442	312	450	322	595	794	2
D	452	313	461	322	595	794	2
Y	461	316	465	323	595	794	2
	430	325	437	332	595	794	2
	459	325	466	332	595	794	2
1	445	327	450	336	595	794	2
D	401	339	410	348	595	794	2
Y	410	342	414	349	595	794	2
=	419	339	427	348	595	794	2
	430	343	437	350	595	794	2
0	445	339	450	347	595	794	2
	459	343	466	350	595	794	2
−1	441	350	454	359	595	794	2
φ	334	366	340	374	595	794	2
=	343	366	350	374	595	794	2
α	353	366	359	374	595	794	2
tan(I	361	366	384	374	595	794	2
Y	384	369	389	375	595	794	2
,	391	366	394	374	595	794	2
I	395	366	400	374	595	794	2
X	400	369	406	375	595	794	2
)	407	366	411	374	595	794	2
(1)	512	339	524	348	595	794	2
(2)	512	366	524	374	595	794	2
Where:	308	386	338	395	595	794	2
I	341	386	345	395	595	794	2
is	348	386	354	395	595	794	2
the	357	386	370	395	595	794	2
image	373	386	398	395	595	794	2
and	401	386	417	395	595	794	2
Ix	420	386	429	396	595	794	2
,	429	386	432	395	595	794	2
Iy	435	386	443	396	595	794	2
are	447	386	460	395	595	794	2
the	463	386	476	395	595	794	2
derivatives	479	386	524	395	595	794	2
of	293	398	301	407	595	794	2
the	304	398	318	407	595	794	2
image	321	398	347	407	595	794	2
in	350	398	358	407	595	794	2
x,y.	362	398	377	407	595	794	2
Each	308	410	329	419	595	794	2
cell	332	410	346	419	595	794	2
is	349	410	356	419	595	794	2
discretized	359	410	404	419	595	794	2
into	408	410	424	419	595	794	2
angular	428	410	460	419	595	794	2
bins	463	410	480	419	595	794	2
according	484	410	524	419	595	794	2
to	293	422	302	431	595	794	2
the	305	422	319	431	595	794	2
gradient	322	422	359	431	595	794	2
orientation	362	422	411	431	595	794	2
and	414	422	430	431	595	794	2
each	434	422	453	431	595	794	2
pixel	456	422	477	431	595	794	2
of	481	422	489	431	595	794	2
the	492	422	506	431	595	794	2
cell	510	422	524	431	595	794	2
contributes	293	434	340	442	595	794	2
with	343	434	362	442	595	794	2
a	365	434	370	442	595	794	2
gradient	372	434	407	442	595	794	2
weight	410	434	438	442	595	794	2
to	440	434	449	442	595	794	2
its	452	434	462	442	595	794	2
corresponding	465	434	524	442	595	794	2
angular	293	446	325	454	595	794	2
bin.	328	446	345	454	595	794	2
The	348	446	365	454	595	794	2
adjacent	368	446	405	454	595	794	2
cells	408	446	426	454	595	794	2
are	429	446	443	454	595	794	2
grouped	446	446	481	454	595	794	2
in	484	446	492	454	595	794	2
special	496	446	524	454	595	794	2
regions	293	458	323	466	595	794	2
called	326	458	351	466	595	794	2
blocks	354	458	381	466	595	794	2
and	384	458	400	466	595	794	2
the	404	458	417	466	595	794	2
normalized	421	458	468	466	595	794	2
group	471	458	496	466	595	794	2
of	499	458	507	466	595	794	2
his-	511	458	526	466	595	794	2
tograms	293	470	328	478	595	794	2
represents	332	470	376	478	595	794	2
the	379	470	394	478	595	794	2
block	397	470	420	478	595	794	2
histogram.	424	470	470	478	595	794	2
Finally,	474	470	507	478	595	794	2
the	510	470	524	478	595	794	2
set	293	481	305	490	595	794	2
of	308	481	316	490	595	794	2
these	319	481	341	490	595	794	2
block	344	481	366	490	595	794	2
histograms	369	481	416	490	595	794	2
represents	419	481	461	490	595	794	2
the	465	481	478	490	595	794	2
descriptor.	481	481	526	490	595	794	2
3.2.	293	507	311	516	595	794	2
Local	316	507	343	516	595	794	2
Binary	347	507	381	516	595	794	2
Patterns	385	507	428	516	595	794	2
(LBP)	431	507	463	516	595	794	2
This	292	525	311	534	595	794	2
algorithm	315	525	356	534	595	794	2
was	359	525	375	534	595	794	2
presented	378	525	419	534	595	794	2
like	422	525	437	534	595	794	2
a	440	525	445	534	595	794	2
texture	448	525	479	534	595	794	2
descriptor	482	525	525	534	595	794	2
for	293	537	305	546	595	794	2
object	309	537	336	546	595	794	2
detection.	341	537	384	546	595	794	2
This	389	537	409	546	595	794	2
compares	413	537	454	546	595	794	2
a	459	537	464	546	595	794	2
central	469	537	499	546	595	794	2
pixel	503	537	524	546	595	794	2
with	292	549	311	558	595	794	2
the	315	549	328	558	595	794	2
neighbors.	331	549	375	558	595	794	2
The	378	549	395	558	595	794	2
central	398	549	427	558	595	794	2
pixel	430	549	451	558	595	794	2
value	454	549	476	558	595	794	2
is	479	549	486	558	595	794	2
taken	489	549	512	558	595	794	2
as	516	549	524	558	595	794	2
threshold	293	561	332	570	595	794	2
and	335	561	351	570	595	794	2
a	354	561	359	570	595	794	2
value	362	561	384	570	595	794	2
of	387	561	395	570	595	794	2
“1”	398	561	412	570	595	794	2
is	415	561	421	570	595	794	2
assigned	424	561	460	570	595	794	2
if	463	561	468	570	595	794	2
the	471	561	485	570	595	794	2
neighbor	488	561	525	570	595	794	2
is	293	573	299	582	595	794	2
greater	303	573	334	582	595	794	2
or	337	573	346	582	595	794	2
equal	349	573	372	582	595	794	2
to	376	573	385	582	595	794	2
the	388	573	402	582	595	794	2
central	405	573	435	582	595	794	2
pixel,	439	573	462	582	595	794	2
otherwise	466	573	507	582	595	794	2
the	510	573	524	582	595	794	2
value	292	585	315	594	595	794	2
is	319	585	326	594	595	794	2
“0”.	330	585	346	594	595	794	2
In	350	585	360	594	595	794	2
each	364	585	383	594	595	794	2
pixel	387	585	408	594	595	794	2
we	412	585	424	594	595	794	2
set	428	585	440	594	595	794	2
a	444	585	449	594	595	794	2
weight	453	585	482	594	595	794	2
of	486	585	494	594	595	794	2
2n	498	585	509	594	595	794	2
ac-	513	585	526	594	595	794	2
cording	293	597	324	606	595	794	2
to	327	597	336	606	595	794	2
the	339	597	352	606	595	794	2
position	355	597	389	606	595	794	2
respect	392	597	422	606	595	794	2
to	425	597	434	606	595	794	2
the	437	597	450	606	595	794	2
central	453	597	482	606	595	794	2
pixel	485	597	506	606	595	794	2
[26].	508	597	526	606	595	794	2
The	292	609	309	618	595	794	2
parameters	312	609	359	618	595	794	2
of	362	609	370	618	595	794	2
LBP	373	609	392	618	595	794	2
operator	395	609	431	618	595	794	2
are	434	609	447	618	595	794	2
R	450	609	457	618	595	794	2
and	460	609	475	618	595	794	2
P,	478	609	487	618	595	794	2
where	489	609	514	618	595	794	2
R	517	609	524	618	595	794	2
is	293	621	299	630	595	794	2
the	302	621	316	630	595	794	2
distance	318	621	353	630	595	794	2
to	356	621	365	630	595	794	2
the	367	621	381	630	595	794	2
central	384	621	413	630	595	794	2
pixel	416	621	436	630	595	794	2
and	439	621	454	630	595	794	2
P	457	621	464	630	595	794	2
is	467	621	473	630	595	794	2
the	476	621	490	630	595	794	2
number	492	621	525	630	595	794	2
of	293	633	301	642	595	794	2
pixels	304	633	329	642	595	794	2
[14].	332	633	350	642	595	794	2
LBP	354	633	374	642	595	794	2
is	377	633	384	642	595	794	2
defined	387	633	418	642	595	794	2
mathematically	422	633	489	642	595	794	2
as:	492	633	504	642	595	794	2
LBP	322	672	343	681	595	794	2
P,R	343	676	357	682	595	794	2
(X	358	672	370	681	595	794	2
C	370	676	376	682	595	794	2
,	377	672	379	681	595	794	2
Y	381	672	387	681	595	794	2
C	387	676	392	682	595	794	2
)	393	672	397	681	595	794	2
=	400	672	408	681	595	794	2
P	411	662	416	668	595	794	2
−1	417	662	427	668	595	794	2
X	412	669	426	676	595	794	2
s(g	429	672	442	681	595	794	2
P	442	676	447	682	595	794	2
−	451	672	458	681	595	794	2
g	461	672	465	681	595	794	2
C	465	676	471	682	595	794	2
)	472	672	476	681	595	794	2
·	478	672	481	681	595	794	2
2	483	672	488	681	595	794	2
P	488	670	493	677	595	794	2
(3)	512	672	524	681	595	794	2
P	411	686	416	693	595	794	2
=0	417	686	427	693	595	794	2
In	308	701	317	710	595	794	2
the	319	701	333	710	595	794	2
equation	335	701	372	710	595	794	2
,	375	701	377	710	595	794	2
gP	380	701	392	714	595	794	2
is	395	701	401	710	595	794	2
the	404	701	417	710	595	794	2
value	420	701	442	710	595	794	2
of	445	701	452	710	595	794	2
the	455	701	469	710	595	794	2
central	471	701	500	710	595	794	2
pixel,	503	701	526	710	595	794	2
gC	293	713	305	726	595	794	2
the	309	713	323	722	595	794	2
value	327	713	350	722	595	794	2
of	354	713	362	722	595	794	2
the	366	713	380	722	595	794	2
neighbors	384	713	427	722	595	794	2
and	431	713	447	722	595	794	2
2P	451	713	463	722	595	794	2
is	467	713	474	722	595	794	2
the	478	713	492	722	595	794	2
weight	496	713	525	722	595	794	2
stablished	293	725	335	734	595	794	2
for	338	725	350	734	595	794	2
each	353	725	371	734	595	794	2
operation.	374	725	418	734	595	794	2
And	421	725	439	734	595	794	2
s(gP	442	725	462	738	595	794	2
−	464	725	472	734	595	794	2
gC	473	725	485	738	595	794	2
)	486	725	490	734	595	794	2
is	493	725	499	734	595	794	2
given	502	725	524	734	595	794	2
by:	293	737	306	746	595	794	2
69	533	29	544	39	595	794	3
Luna	74	30	96	39	595	794	3
et	100	30	108	39	595	794	3
al.	112	30	122	39	595	794	3
/	124	30	129	39	595	794	3
Mobile	133	30	161	39	595	794	3
robot	165	30	186	39	595	794	3
with	190	30	208	39	595	794	3
vision	211	30	237	39	595	794	3
based	240	30	263	39	595	794	3
navigation	266	30	311	39	595	794	3
and	315	30	331	39	595	794	3
pedestrian	334	30	378	39	595	794	3
detection	382	30	420	39	595	794	3
s(g	112	82	125	91	595	794	3
P	125	86	130	92	595	794	3
−	134	82	141	91	595	794	3
g	144	82	148	91	595	794	3
C	148	86	154	92	595	794	3
)	155	82	159	91	595	794	3
=	162	82	170	91	595	794	3
1	185	76	190	85	595	794	3
0	185	88	190	97	595	794	3
g	200	76	204	85	595	794	3
P	204	80	209	86	595	794	3
−	213	76	221	85	595	794	3
g	223	76	228	85	595	794	3
C	228	80	234	86	595	794	3
≥	237	76	245	85	595	794	3
0;	248	76	256	85	595	794	3
g	200	88	204	97	595	794	3
P	204	92	209	98	595	794	3
−	213	88	221	97	595	794	3
g	223	88	228	97	595	794	3
C	228	92	234	98	595	794	3
0	248	88	253	97	595	794	3
cot	387	81	400	90	595	794	3
θ	403	81	408	90	595	794	3
i	408	85	411	91	595	794	3
−	414	81	421	90	595	794	3
cot	423	81	436	90	595	794	3
θ	440	81	444	90	595	794	3
o	444	85	448	91	595	794	3
=	451	82	459	90	595	794	3
(4)	290	82	303	91	595	794	3
d	463	75	468	84	595	794	3
l	464	88	467	97	595	794	3
(6)	532	82	544	90	595	794	3
where:	327	100	356	108	595	794	3
3.3.	71	115	89	123	595	794	3
Adaboost	94	115	142	123	595	794	3
Adaboost	70	133	113	142	595	794	3
is	117	133	124	142	595	794	3
a	128	133	133	142	595	794	3
machine	137	133	174	142	595	794	3
learning	178	133	213	142	595	794	3
algorithm	217	133	260	142	595	794	3
[33]	264	133	280	142	595	794	3
that	284	133	303	142	595	794	3
initially	71	145	104	154	595	794	3
keeps	108	145	131	154	595	794	3
uniform	134	145	168	154	595	794	3
distribution	171	145	222	154	595	794	3
of	225	145	233	154	595	794	3
weights	237	145	269	154	595	794	3
in	272	145	280	154	595	794	3
each	284	145	303	154	595	794	3
training	71	157	106	166	595	794	3
sample.	110	157	143	166	595	794	3
In	147	157	156	166	595	794	3
the	160	157	174	166	595	794	3
first	178	157	196	166	595	794	3
iteration	200	157	238	166	595	794	3
the	242	157	256	166	595	794	3
algorithm	260	157	303	166	595	794	3
trains	71	169	95	178	595	794	3
a	98	169	103	178	595	794	3
weak	106	169	127	178	595	794	3
classifier	130	169	166	178	595	794	3
using	168	169	191	178	595	794	3
a	194	169	198	178	595	794	3
feature	201	169	231	178	595	794	3
extraction	233	169	277	178	595	794	3
meth-	279	169	304	178	595	794	3
ods	71	181	86	190	595	794	3
or	91	181	100	190	595	794	3
mix	104	181	121	190	595	794	3
of	126	181	134	190	595	794	3
them	138	181	161	190	595	794	3
achieving	166	181	207	190	595	794	3
a	211	181	216	190	595	794	3
higher	221	181	249	190	595	794	3
recognition	253	181	303	190	595	794	3
performance	71	193	125	202	595	794	3
for	128	193	140	202	595	794	3
the	143	193	157	202	595	794	3
training	160	193	195	202	595	794	3
samples.	198	193	235	202	595	794	3
In	86	205	95	214	595	794	3
the	98	205	112	214	595	794	3
second	115	205	144	214	595	794	3
iteration,	148	205	188	214	595	794	3
the	191	205	205	214	595	794	3
training	208	205	243	214	595	794	3
samples,	246	205	283	214	595	794	3
mis-	286	205	304	214	595	794	3
classified	71	217	110	226	595	794	3
by	114	217	125	226	595	794	3
the	129	217	144	226	595	794	3
first	148	217	166	226	595	794	3
weak	170	217	192	226	595	794	3
classifier,	196	217	237	226	595	794	3
receive	241	217	271	226	595	794	3
higher	275	217	303	226	595	794	3
weights.	71	229	105	237	595	794	3
The	86	241	103	249	595	794	3
new	105	241	122	249	595	794	3
selected	124	241	157	249	595	794	3
feature	159	241	189	249	595	794	3
extraction	191	241	234	249	595	794	3
methods	236	241	273	249	595	794	3
should	275	241	303	249	595	794	3
be	71	253	81	261	595	794	3
focused	85	253	118	261	595	794	3
in	122	253	131	261	595	794	3
these	135	253	157	261	595	794	3
misclassified	161	253	216	261	595	794	3
samples.	220	253	257	261	595	794	3
The	262	253	279	261	595	794	3
final	283	253	303	261	595	794	3
result	71	265	95	273	595	794	3
is	98	265	104	273	595	794	3
a	107	265	112	273	595	794	3
cascade	115	265	147	273	595	794	3
of	150	265	158	273	595	794	3
linear	161	265	185	273	595	794	3
combinations	188	265	244	273	595	794	3
with	247	265	266	273	595	794	3
selected	269	265	303	273	595	794	3
weak	71	276	92	285	595	794	3
classifiers:	95	276	139	285	595	794	3
g(x)	147	308	165	316	595	794	3
=	168	308	176	316	595	794	3
T	183	297	187	303	595	794	3
X	178	305	193	311	595	794	3
℘	194	308	201	316	595	794	3
t	201	306	204	312	595	794	3
h	204	308	210	316	595	794	3
t	210	306	213	312	595	794	3
(x)	213	308	227	316	595	794	3
(5)	290	308	303	316	595	794	3
θ	327	123	333	132	595	794	3
i	333	127	336	133	595	794	3
=	336	123	344	132	595	794	3
Relative	347	123	383	132	595	794	3
angle	386	123	409	132	595	794	3
of	412	123	420	132	595	794	3
the	424	123	437	132	595	794	3
inner	441	123	463	132	595	794	3
wheel.	466	123	493	132	595	794	3
θ	327	135	333	144	595	794	3
o	333	139	337	145	595	794	3
=	337	135	345	144	595	794	3
Relative	349	135	384	144	595	794	3
angle	387	135	410	144	595	794	3
of	413	135	421	144	595	794	3
an	425	135	435	144	595	794	3
outer	439	135	461	144	595	794	3
wheel.	465	135	492	144	595	794	3
l=	327	147	338	156	595	794	3
Longitudinal	341	147	397	156	595	794	3
separation	400	147	445	156	595	794	3
between	448	147	484	156	595	794	3
wheels.	487	147	518	156	595	794	3
d=	327	159	341	168	595	794	3
Lateral	344	159	375	168	595	794	3
separation	379	159	423	168	595	794	3
between	427	159	462	168	595	794	3
wheels.	465	159	496	168	595	794	3
For	327	183	342	192	595	794	3
traction	346	183	381	192	595	794	3
and	385	183	401	192	595	794	3
steering	405	183	440	192	595	794	3
control	443	183	474	192	595	794	3
the	478	183	492	192	595	794	3
system	496	183	526	192	595	794	3
use	530	183	544	192	595	794	3
Pulse	313	195	336	204	595	794	3
Width	340	195	369	204	595	794	3
Modulator	373	195	420	204	595	794	3
(PWM)	423	195	458	204	595	794	3
with	462	195	481	204	595	794	3
Chopper	485	195	523	204	595	794	3
con-	527	195	546	204	595	794	3
figuration	313	207	356	216	595	794	3
[34].	359	207	378	216	595	794	3
Additionally,	382	207	439	216	595	794	3
there	442	207	465	216	595	794	3
is	469	207	475	216	595	794	3
a	479	207	484	216	595	794	3
smart	488	207	513	216	595	794	3
device	517	207	544	216	595	794	3
(smartphone)	311	219	370	228	595	794	3
camera	373	219	404	228	595	794	3
for	408	219	420	228	595	794	3
navigation.	423	219	471	228	595	794	3
The	327	231	344	240	595	794	3
robot	347	231	371	240	595	794	3
has	374	231	388	240	595	794	3
two	391	231	406	240	595	794	3
motors,	409	231	442	240	595	794	3
one	445	231	459	240	595	794	3
for	463	231	474	240	595	794	3
forward/reverse	477	231	544	240	595	794	3
displacement	313	243	369	252	595	794	3
and	372	243	388	252	595	794	3
other	391	243	414	252	595	794	3
that	417	243	436	252	595	794	3
controls	439	243	473	252	595	794	3
direction.	476	243	517	252	595	794	3
An	520	243	533	252	595	794	3
H	537	243	544	252	595	794	3
bridges	313	255	344	264	595	794	3
circuit	347	255	374	264	595	794	3
for	378	255	390	264	595	794	3
motor	393	255	419	264	595	794	3
control	422	255	452	264	595	794	3
is	456	255	462	264	595	794	3
implemented.	466	255	524	264	595	794	3
The	527	255	544	264	595	794	3
camera	313	267	343	276	595	794	3
of	347	267	355	276	595	794	3
smartphone	358	267	408	276	595	794	3
is	412	267	418	276	595	794	3
used	422	267	441	276	595	794	3
as	444	267	453	276	595	794	3
IP	456	267	467	276	595	794	3
camera	470	267	501	276	595	794	3
for	504	267	516	276	595	794	3
image	519	267	544	276	595	794	3
capture.	313	279	348	288	595	794	3
4.2.	313	304	330	313	595	794	3
Pedestrian	336	304	389	313	595	794	3
Detection	393	304	442	313	595	794	3
t=1	179	321	192	328	595	794	3
The	312	322	329	331	595	794	3
feature	332	322	361	331	595	794	3
extraction	364	322	407	331	595	794	3
algorithms	410	322	455	331	595	794	3
previously	458	322	501	331	595	794	3
described	504	322	544	331	595	794	3
are	313	334	326	343	595	794	3
applied	331	334	364	343	595	794	3
on	369	334	380	343	595	794	3
the	385	334	399	343	595	794	3
mobile	404	334	434	343	595	794	3
robot	439	334	463	343	595	794	3
to	468	334	477	343	595	794	3
obtain	482	334	511	343	595	794	3
an	516	334	527	343	595	794	3
au-	532	334	546	343	595	794	3
tonomous	313	346	354	355	595	794	3
navigation	356	346	401	355	595	794	3
system	403	346	432	355	595	794	3
based	435	346	459	355	595	794	3
on	461	346	472	355	595	794	3
pedestrian	474	346	518	355	595	794	3
track-	521	346	546	355	595	794	3
ing.	313	358	329	367	595	794	3
We	333	358	347	367	595	794	3
use	351	358	366	367	595	794	3
Adaboost	370	358	412	367	595	794	3
as	416	358	425	367	595	794	3
machine	429	358	466	367	595	794	3
learning	470	358	506	367	595	794	3
method.	510	358	546	367	595	794	3
4.	71	364	81	374	595	794	3
Our	88	364	111	374	595	794	3
Approach	116	364	173	374	595	794	3
It	313	370	320	379	595	794	3
is	324	370	331	379	595	794	3
an	334	370	345	379	595	794	3
algorithm	349	370	392	379	595	794	3
in	395	370	404	379	595	794	3
machine	408	370	444	379	595	794	3
learning	448	370	483	379	595	794	3
based	487	370	512	379	595	794	3
on	516	370	526	379	595	794	3
the	530	370	544	379	595	794	3
notion	313	382	341	391	595	794	3
of	345	382	353	391	595	794	3
creating	357	382	393	391	595	794	3
a	397	382	402	391	595	794	3
highly	406	382	433	391	595	794	3
accurate	437	382	475	391	595	794	3
prediction	479	382	523	391	595	794	3
rule	527	382	544	391	595	794	3
4.1.	71	388	89	397	595	794	3
Mobile	94	388	129	397	595	794	3
Robotic	133	388	172	397	595	794	3
Platform	176	388	221	397	595	794	3
by	313	394	323	403	595	794	3
combining	328	394	373	403	595	794	3
many	377	394	402	403	595	794	3
relatively	406	394	447	403	595	794	3
weak	451	394	473	403	595	794	3
and	478	394	494	403	595	794	3
inaccurate	498	394	544	403	595	794	3
We	70	407	84	415	595	794	3
are	87	407	101	415	595	794	3
using	104	407	127	415	595	794	3
the	130	407	144	415	595	794	3
Ackerman	147	407	190	415	595	794	3
steering	194	407	227	415	595	794	3
configuration	231	407	288	415	595	794	3
for	291	407	303	415	595	794	3
rules	313	406	333	415	595	794	3
[33].	336	406	355	415	595	794	3
Fig.	358	406	375	415	595	794	3
2	378	406	383	415	595	794	3
shows	386	406	412	415	595	794	3
images	415	406	444	415	595	794	3
of	448	406	456	415	595	794	3
the	459	406	473	415	595	794	3
HOG	476	406	499	415	595	794	3
algorithm	502	406	544	415	595	794	3
the	71	419	85	427	595	794	3
robot	88	419	112	427	595	794	3
[19],	116	419	134	427	595	794	3
i.e.	138	419	151	427	595	794	3
two	154	419	170	427	595	794	3
wheels	173	419	202	427	595	794	3
with	205	419	225	427	595	794	3
rear	228	419	246	427	595	794	3
haulage	249	419	283	427	595	794	3
and	286	419	303	427	595	794	3
performance	313	418	366	427	595	794	3
in	370	418	378	427	595	794	3
the	381	418	395	427	595	794	3
mobile	398	418	427	427	595	794	3
robot.	430	418	457	427	595	794	3
two	71	431	87	439	595	794	3
with	90	431	109	439	595	794	3
front	113	431	134	439	595	794	3
direction	137	431	176	439	595	794	3
presented	179	431	221	439	595	794	3
in	225	431	233	439	595	794	3
the	236	431	250	439	595	794	3
Fig.	254	431	271	439	595	794	3
1.	274	431	282	439	595	794	3
The	285	431	303	439	595	794	3
system	71	443	101	451	595	794	3
is	104	443	111	451	595	794	3
designed	114	443	152	451	595	794	3
so	155	443	164	451	595	794	3
that	167	443	186	451	595	794	3
the	189	443	203	451	595	794	3
interior	206	443	238	451	595	794	3
front	242	443	263	451	595	794	3
wheel	266	443	291	451	595	794	3
in	294	443	303	451	595	794	3
a	71	454	76	463	595	794	3
gyre	79	454	97	463	595	794	3
has	101	454	115	463	595	794	3
an	118	454	129	463	595	794	3
angle	132	454	155	463	595	794	3
slightly	158	454	189	463	595	794	3
sharper	193	454	225	463	595	794	3
than	228	454	248	463	595	794	3
the	251	454	265	463	595	794	3
exterior,	268	454	304	463	595	794	3
in	71	466	79	475	595	794	3
order	82	466	104	475	595	794	3
to	107	466	116	475	595	794	3
avoid	119	466	142	475	595	794	3
skidding;	145	466	183	475	595	794	3
the	186	466	200	475	595	794	3
normal	203	466	233	475	595	794	3
wheels	236	466	264	475	595	794	3
intersect	267	466	303	475	595	794	3
at	71	478	80	487	595	794	3
a	83	478	88	487	595	794	3
point	91	478	114	487	595	794	3
on	118	478	128	487	595	794	3
extension	132	478	172	487	595	794	3
of	176	478	184	487	595	794	3
the	187	478	201	487	595	794	3
axis	204	478	222	487	595	794	3
of	225	478	233	487	595	794	3
the	236	478	250	487	595	794	3
rear	253	478	271	487	595	794	3
wheels	274	478	303	487	595	794	3
and	71	490	87	499	595	794	3
make	90	490	113	499	595	794	3
the	116	490	130	499	595	794	3
toolpath	133	490	170	499	595	794	3
for	174	490	185	499	595	794	3
constant	189	490	226	499	595	794	3
rotation	229	490	264	499	595	794	3
angles.	267	490	297	499	595	794	3
where:	71	336	99	345	595	794	3
ht	102	332	111	345	595	794	3
is	115	336	122	345	595	794	3
a	125	336	130	345	595	794	3
classifier,	134	336	173	345	595	794	3
and	177	336	193	345	595	794	3
℘	196	336	202	345	595	794	3
t	202	333	206	342	595	794	3
is	210	336	217	345	595	794	3
a	220	336	225	345	595	794	3
coefficient.	228	336	275	345	595	794	3
Figure	313	536	343	544	595	794	3
2.	346	536	354	544	595	794	3
Pedestrian	358	536	400	544	595	794	3
detection	404	536	440	544	595	794	3
with	444	536	461	544	595	794	3
HOG	465	536	486	544	595	794	3
algorithm.	489	536	530	544	595	794	3
(a)	534	536	545	544	595	794	3
Side	313	547	329	555	595	794	3
detection	332	547	369	555	595	794	3
of	372	547	379	555	595	794	3
a	382	547	387	555	595	794	3
person.	390	547	419	555	595	794	3
(b)	422	547	434	555	595	794	3
Back	437	547	457	555	595	794	3
detection	460	547	496	555	595	794	3
of	499	547	507	555	595	794	3
a	510	547	514	555	595	794	3
person.	517	547	546	555	595	794	3
(c)	311	558	323	566	595	794	3
Frontal	326	558	355	566	595	794	3
Detection	358	558	397	566	595	794	3
of	400	558	407	566	595	794	3
a	410	558	415	566	595	794	3
person.	418	558	447	566	595	794	3
Performance	327	577	381	586	595	794	3
of	384	577	392	586	595	794	3
LBP	395	577	415	586	595	794	3
algorithm	418	577	460	586	595	794	3
in	463	577	471	586	595	794	3
images	475	577	503	586	595	794	3
captured	507	577	544	586	595	794	3
from	313	589	333	598	595	794	3
the	336	589	350	598	595	794	3
mobile	353	589	382	598	595	794	3
robot	385	589	409	598	595	794	3
is	412	589	419	598	595	794	3
presented	422	589	464	598	595	794	3
on	467	589	477	598	595	794	3
Fig.	481	589	498	598	595	794	3
3.	501	589	509	598	595	794	3
Figure	71	694	101	702	595	794	3
1.	104	694	113	702	595	794	3
Position	116	694	148	702	595	794	3
of	151	694	159	702	595	794	3
wheels	162	694	188	702	595	794	3
in	191	694	198	702	595	794	3
Ackerman	201	694	241	702	595	794	3
steering	244	694	275	702	595	794	3
config-	278	694	304	702	595	794	3
Figure	313	704	343	712	595	794	3
3.	347	704	356	712	595	794	3
Fig.	360	704	376	712	595	794	3
3.	380	704	388	712	595	794	3
Pedestrian	392	704	435	712	595	794	3
detection	439	704	477	712	595	794	3
with	481	704	499	712	595	794	3
LBP	503	704	522	712	595	794	3
algo-	526	704	546	712	595	794	3
uration.	71	705	103	713	595	794	3
rithm.	313	715	338	723	595	794	3
(a)	342	715	354	723	595	794	3
Detection	357	715	396	723	595	794	3
of	400	715	407	723	595	794	3
a	411	715	415	723	595	794	3
false	419	715	437	723	595	794	3
positive.	440	715	474	723	595	794	3
(b)	478	715	490	723	595	794	3
Detection	494	715	533	723	595	794	3
of	537	715	544	723	595	794	3
The	86	725	103	734	595	794	3
relationship	106	725	157	734	595	794	3
between	160	725	196	734	595	794	3
the	199	725	213	734	595	794	3
angles	216	725	243	734	595	794	3
of	246	725	254	734	595	794	3
the	257	725	271	734	595	794	3
wheels	274	725	303	734	595	794	3
two	313	726	327	734	595	794	3
people.	330	726	358	734	595	794	3
(c)	361	726	373	734	595	794	3
Frontal	376	726	405	734	595	794	3
detection	408	726	445	734	595	794	3
of	448	726	455	734	595	794	3
a	458	726	463	734	595	794	3
person.	466	726	495	734	595	794	3
of	71	737	79	746	595	794	3
direction	82	737	120	746	595	794	3
is	124	737	130	746	595	794	3
in	134	737	142	746	595	794	3
equation	145	737	183	746	595	794	3
6:	186	737	194	746	595	794	3
INGENIUS	337	30	396	39	595	794	4
N.	399	30	409	39	595	794	4
◦	409	28	413	35	595	794	4
17,	417	30	430	39	595	794	4
Enero-Junio	433	30	486	39	595	794	4
de	491	30	501	39	595	794	4
2017	504	30	524	39	595	794	4
70	51	29	62	39	595	794	4
4.3.	51	67	69	76	595	794	4
Controller	75	67	126	76	595	794	4
1000	292	67	312	76	595	794	4
frames	315	67	343	76	595	794	4
separated	346	67	386	76	595	794	4
in	389	67	397	76	595	794	4
groups	400	67	429	76	595	794	4
of	431	67	439	76	595	794	4
30	442	67	452	76	595	794	4
frames.	455	67	486	76	595	794	4
For	489	67	503	76	595	794	4
each	506	67	524	76	595	794	4
group,	293	79	321	88	595	794	4
we	325	79	336	88	595	794	4
determined	340	79	390	88	595	794	4
true	394	79	412	88	595	794	4
positives,	416	79	457	88	595	794	4
true	460	79	478	88	595	794	4
negatives,	482	79	526	88	595	794	4
false	293	91	311	100	595	794	4
positives	314	91	351	100	595	794	4
and	354	91	370	100	595	794	4
false	372	91	391	100	595	794	4
negatives	394	91	433	100	595	794	4
using	436	91	458	100	595	794	4
HOG	461	91	483	100	595	794	4
and	486	91	502	100	595	794	4
LBP	505	91	524	100	595	794	4
algorithms.	293	103	341	112	595	794	4
The	345	103	362	112	595	794	4
image	365	103	390	112	595	794	4
processing	394	103	438	112	595	794	4
was	441	103	457	112	595	794	4
realized	460	103	494	112	595	794	4
with	497	103	516	112	595	794	4
a	519	103	524	112	595	794	4
2.4	292	115	305	123	595	794	4
GHz	308	115	328	123	595	794	4
processor	331	115	371	123	595	794	4
and	374	115	390	123	595	794	4
RAM	393	115	417	123	595	794	4
memory	420	115	455	123	595	794	4
of	458	115	466	123	595	794	4
4	469	115	474	123	595	794	4
GB.	478	115	495	123	595	794	4
In	498	115	507	123	595	794	4
the	511	115	524	123	595	794	4
online	293	127	319	135	595	794	4
test	322	127	338	135	595	794	4
we	342	127	353	135	595	794	4
recorded	356	127	394	135	595	794	4
7	397	127	402	135	595	794	4
minutes	405	127	440	135	595	794	4
of	443	127	451	135	595	794	4
video,	454	127	480	135	595	794	4
i.e.	484	127	496	135	595	794	4
12600	500	127	525	135	595	794	4
frames,	293	139	325	147	595	794	4
and	329	139	345	147	595	794	4
the	350	139	364	147	595	794	4
same	368	139	390	147	595	794	4
procedure	395	139	439	147	595	794	4
in	443	139	452	147	595	794	4
offline	456	139	483	147	595	794	4
test	487	139	504	147	595	794	4
was	508	139	524	147	595	794	4
performed.	293	151	341	159	595	794	4
The	344	151	362	159	595	794	4
results	365	151	394	159	595	794	4
are	397	151	411	159	595	794	4
presented	414	151	456	159	595	794	4
in	460	151	468	159	595	794	4
Table	472	151	496	159	595	794	4
1	500	151	505	159	595	794	4
and	508	151	524	159	595	794	4
Table	292	163	316	171	595	794	4
2.	320	163	327	171	595	794	4
The	308	174	325	183	595	794	4
Table	329	174	354	183	595	794	4
1	358	174	363	183	595	794	4
presents	367	174	403	183	595	794	4
the	407	174	421	183	595	794	4
results	425	174	454	183	595	794	4
of	458	174	466	183	595	794	4
recall	470	174	494	183	595	794	4
of	498	174	506	183	595	794	4
the	510	174	524	183	595	794	4
detectors	293	186	332	195	595	794	4
tested	335	186	361	195	595	794	4
offline	365	186	391	195	595	794	4
and	395	186	411	195	595	794	4
online.	414	186	443	195	595	794	4
The	51	85	68	94	595	794	4
mobile	72	85	101	94	595	794	4
robot	105	85	129	94	595	794	4
uses	133	85	151	94	595	794	4
two	155	85	171	94	595	794	4
main	175	85	197	94	595	794	4
inputs	200	85	228	94	595	794	4
for	232	85	244	94	595	794	4
the	248	85	262	94	595	794	4
con-	266	85	284	94	595	794	4
troller	51	97	78	106	595	794	4
(needed	81	97	115	106	595	794	4
for	118	97	130	106	595	794	4
2D	133	97	146	106	595	794	4
navigation	149	97	195	106	595	794	4
[35]):	198	97	220	106	595	794	4
horizontal	223	97	267	106	595	794	4
po-	270	97	284	106	595	794	4
sition	51	109	74	118	595	794	4
and	77	109	93	118	595	794	4
distance	96	109	131	118	595	794	4
obtained	134	109	171	118	595	794	4
from	174	109	194	118	595	794	4
the	197	109	210	118	595	794	4
bounding	213	109	253	118	595	794	4
box	256	109	272	118	595	794	4
of	275	109	283	118	595	794	4
the	51	121	65	130	595	794	4
pedestrian	68	121	113	130	595	794	4
detection.	116	121	158	130	595	794	4
Horizontal	162	121	207	130	595	794	4
position	210	121	245	130	595	794	4
depends	248	121	283	130	595	794	4
of	51	133	59	142	595	794	4
x	62	133	68	142	595	794	4
coordinate	71	133	116	142	595	794	4
of	120	133	128	142	595	794	4
the	131	133	145	142	595	794	4
bounding	148	133	189	142	595	794	4
box	192	133	208	142	595	794	4
centroid	211	133	246	142	595	794	4
and	250	133	266	142	595	794	4
the	269	133	283	142	595	794	4
distance	51	145	87	154	595	794	4
depends	91	145	127	154	595	794	4
of	131	145	139	154	595	794	4
the	143	145	157	154	595	794	4
width	161	145	186	154	595	794	4
of	190	145	198	154	595	794	4
the	202	145	216	154	595	794	4
bounding	220	145	262	154	595	794	4
box.	266	145	285	154	595	794	4
We	51	157	65	166	595	794	4
use	68	157	82	166	595	794	4
PI	85	157	96	166	595	794	4
control	99	157	130	166	595	794	4
for	133	157	145	166	595	794	4
the	148	157	162	166	595	794	4
distance	166	157	202	166	595	794	4
that	205	157	224	166	595	794	4
reduces	227	157	259	166	595	794	4
over-	263	157	284	166	595	794	4
shoot	51	169	74	178	595	794	4
in	77	169	86	178	595	794	4
order	89	169	111	178	595	794	4
to	115	169	123	178	595	794	4
have	127	169	146	178	595	794	4
a	149	169	154	178	595	794	4
soft	157	169	173	178	595	794	4
controller.	176	169	219	178	595	794	4
For	223	169	237	178	595	794	4
horizontal	240	169	283	178	595	794	4
position	51	181	85	190	595	794	4
we	88	181	99	190	595	794	4
use	102	181	116	190	595	794	4
an	119	181	129	190	595	794	4
on-off	132	181	157	190	595	794	4
controller	160	181	200	190	595	794	4
with	203	181	222	190	595	794	4
hysteresis,	226	181	269	190	595	794	4
i.e.	272	181	285	190	595	794	4
with	51	193	70	202	595	794	4
a	73	193	78	202	595	794	4
band	82	193	103	202	595	794	4
of	106	193	114	202	595	794	4
maximum	118	193	161	202	595	794	4
and	164	193	180	202	595	794	4
minimum	183	193	224	202	595	794	4
values	228	193	254	202	595	794	4
in	257	193	266	202	595	794	4
the	269	193	283	202	595	794	4
output	51	205	81	214	595	794	4
due	84	205	100	214	595	794	4
to	103	205	112	214	595	794	4
the	115	205	130	214	595	794	4
variation	133	205	172	214	595	794	4
range	175	205	199	214	595	794	4
of	203	205	211	214	595	794	4
the	214	205	228	214	595	794	4
actuators	232	205	273	214	595	794	4
is	276	205	283	214	595	794	4
Table	292	207	318	215	595	794	4
1.	323	207	331	215	595	794	4
Detection	336	208	375	215	595	794	4
recall	380	208	402	215	595	794	4
of	406	208	414	215	595	794	4
the	418	208	431	215	595	794	4
algorithms	435	208	479	215	595	794	4
previously	483	208	525	215	595	794	4
small	51	217	74	226	595	794	4
and	77	217	93	226	595	794	4
precision	96	217	135	226	595	794	4
is	138	217	145	226	595	794	4
not	148	217	162	226	595	794	4
necessary	166	217	207	226	595	794	4
(Fig.	210	217	231	226	595	794	4
4).	234	217	246	226	595	794	4
trained	293	219	321	226	595	794	4
Algorithm	324	243	376	252	595	794	4
Recall	437	243	468	252	595	794	4
Offline	412	255	447	264	595	794	4
Online	459	255	492	264	595	794	4
HOG	324	272	347	281	595	794	4
+Adaboost	351	272	400	281	595	794	4
LBP+Adaboost	324	284	394	293	595	794	4
78%	420	272	439	281	595	794	4
76,50%	414	284	445	293	595	794	4
74,20%	460	272	491	281	595	794	4
75%	467	284	485	293	595	794	4
Table	292	323	318	331	595	794	4
2.	321	323	329	331	595	794	4
Detection	333	323	371	331	595	794	4
precision	375	323	410	331	595	794	4
of	413	323	420	331	595	794	4
the	423	323	436	331	595	794	4
algorithms	439	323	481	331	595	794	4
previously	484	323	525	331	595	794	4
trained.	293	334	324	342	595	794	4
Figure	77	328	107	336	595	794	4
4.	110	328	118	336	595	794	4
Robot	122	328	147	336	595	794	4
navigation	150	328	192	336	595	794	4
control	195	328	223	336	595	794	4
scheme.	226	328	257	336	595	794	4
Algorithm	324	358	375	367	595	794	4
4.4.	51	365	69	374	595	794	4
Communication	75	365	154	374	595	794	4
Accuracy	431	358	477	367	595	794	4
Offline	415	370	448	379	595	794	4
Online	460	370	493	379	595	794	4
Wi-Fi	51	384	77	392	595	794	4
communication	81	384	149	392	595	794	4
is	153	384	160	392	595	794	4
implemented	165	384	222	392	595	794	4
for	226	384	238	392	595	794	4
transmit-	243	384	284	392	595	794	4
HOG	324	387	347	396	595	794	4
+	350	387	358	396	595	794	4
Adaboost	361	387	403	396	595	794	4
76%	422	387	441	396	595	794	4
74%	468	387	486	396	595	794	4
ting	51	396	69	404	595	794	4
data	72	396	92	404	595	794	4
from	95	396	116	404	595	794	4
the	119	396	133	404	595	794	4
IP	136	396	147	404	595	794	4
camera	150	396	182	404	595	794	4
to	185	396	194	404	595	794	4
the	198	396	212	404	595	794	4
computer.	215	396	260	404	595	794	4
This	263	396	283	404	595	794	4
LBP	324	399	344	408	595	794	4
+	347	399	355	408	595	794	4
Adaboost	358	399	400	408	595	794	4
73%	422	399	441	408	595	794	4
72%	468	399	486	408	595	794	4
is	51	408	58	416	595	794	4
because	62	408	96	416	595	794	4
Wi-Fi	100	408	126	416	595	794	4
is	130	408	137	416	595	794	4
able	141	408	159	416	595	794	4
to	162	408	172	416	595	794	4
send	175	408	195	416	595	794	4
image	199	408	225	416	595	794	4
data.	229	408	252	416	595	794	4
In	255	408	265	416	595	794	4
the	269	408	283	416	595	794	4
other	51	419	74	428	595	794	4
side	78	419	95	428	595	794	4
we	98	419	110	428	595	794	4
use	114	419	128	428	595	794	4
Bluetooth	131	419	176	428	595	794	4
communication	179	419	247	428	595	794	4
to	250	419	259	428	595	794	4
send	263	419	283	428	595	794	4
The	308	425	325	434	595	794	4
HOG	329	425	353	434	595	794	4
precision	357	425	396	434	595	794	4
test	400	425	417	434	595	794	4
is	421	425	428	434	595	794	4
better	432	425	459	434	595	794	4
than	463	425	483	434	595	794	4
LBP	488	425	508	434	595	794	4
be-	512	425	526	434	595	794	4
control	51	431	81	440	595	794	4
command	83	431	125	440	595	794	4
from	127	431	147	440	595	794	4
the	150	431	164	440	595	794	4
computer	166	431	207	440	595	794	4
to	209	431	218	440	595	794	4
the	221	431	234	440	595	794	4
robot.	237	431	263	440	595	794	4
The	266	431	283	440	595	794	4
cause	293	437	316	446	595	794	4
LBP	320	437	341	446	595	794	4
performance	345	437	399	446	595	794	4
give	403	437	421	446	595	794	4
more	425	437	447	446	595	794	4
false	451	437	470	446	595	794	4
positives	474	437	512	446	595	794	4
in	516	437	524	446	595	794	4
transmission	51	443	106	452	595	794	4
speed	109	443	133	452	595	794	4
obtained	137	443	175	452	595	794	4
from	178	443	198	452	595	794	4
experimentation	202	443	273	452	595	794	4
is	276	443	283	452	595	794	4
both	293	449	313	458	595	794	4
cases	317	449	339	458	595	794	4
(offline	343	449	374	458	595	794	4
and	378	449	395	458	595	794	4
online).	398	449	432	458	595	794	4
The	436	449	453	458	595	794	4
LBP	457	449	478	458	595	794	4
algorithm	481	449	524	458	595	794	4
115200	51	455	80	464	595	794	4
baud.	84	455	108	464	595	794	4
(See	111	455	130	464	595	794	4
section	133	455	163	464	595	794	4
5).	166	455	178	464	595	794	4
compares	293	461	334	470	595	794	4
image	338	461	364	470	595	794	4
textures	367	461	403	470	595	794	4
and	407	461	423	470	595	794	4
confuses	426	461	463	470	595	794	4
other	467	461	490	470	595	794	4
objects	493	461	524	470	595	794	4
with	292	473	312	482	595	794	4
persons	315	473	347	482	595	794	4
easily.	351	473	377	482	595	794	4
HOG	380	473	403	482	595	794	4
is	406	473	413	482	595	794	4
a	416	473	421	482	595	794	4
descriptor	424	473	468	482	595	794	4
based	471	473	495	482	595	794	4
on	499	473	509	482	595	794	4
ob-	512	473	526	482	595	794	4
5.	51	483	61	493	595	794	4
Experimentation	68	483	167	493	595	794	4
and	171	483	193	493	595	794	4
Results	197	483	240	493	595	794	4
jects	293	485	312	494	595	794	4
shape	315	485	339	494	595	794	4
and	342	485	358	494	595	794	4
focused	361	485	392	494	595	794	4
principally	395	485	441	494	595	794	4
on	444	485	454	494	595	794	4
pedestrians	457	485	505	494	595	794	4
[24],	508	485	526	494	595	794	4
thus	293	497	312	506	595	794	4
it	315	497	322	506	595	794	4
has	325	497	340	506	595	794	4
better	343	497	370	506	595	794	4
performance	374	497	428	506	595	794	4
but	432	497	447	506	595	794	4
with	450	497	470	506	595	794	4
higher	474	497	501	506	595	794	4
com-	505	497	526	506	595	794	4
5.1.	51	508	69	516	595	794	4
Pedestrian	75	508	128	516	595	794	4
Detection	132	508	180	516	595	794	4
putational	293	509	337	518	595	794	4
cost.	340	509	360	518	595	794	4
Based	363	509	388	518	595	794	4
on	392	509	402	518	595	794	4
the	405	509	419	518	595	794	4
results,	422	509	453	518	595	794	4
we	456	509	467	518	595	794	4
choose	471	509	498	518	595	794	4
HOG	502	509	524	518	595	794	4
because	293	521	327	529	595	794	4
the	331	521	345	529	595	794	4
computational	350	521	413	529	595	794	4
cost	418	521	435	529	595	794	4
problem	439	521	476	529	595	794	4
can	480	521	495	529	595	794	4
be	499	521	510	529	595	794	4
re-	514	521	526	529	595	794	4
The	51	526	68	535	595	794	4
performance	72	526	127	535	595	794	4
of	130	526	138	535	595	794	4
pedestrian	142	526	188	535	595	794	4
detection	192	526	232	535	595	794	4
algorithms	236	526	283	535	595	794	4
duced	293	533	318	541	595	794	4
using	321	533	344	541	595	794	4
a	348	533	353	541	595	794	4
processor	356	533	396	541	595	794	4
with	399	533	419	541	595	794	4
better	422	533	448	541	595	794	4
characteristics.	452	533	516	541	595	794	4
was	51	538	66	547	595	794	4
compared	70	538	112	547	595	794	4
using	115	538	138	547	595	794	4
online	141	538	167	547	595	794	4
and	170	538	186	547	595	794	4
offline	190	538	216	547	595	794	4
videos.	219	538	249	547	595	794	4
We	252	538	266	547	595	794	4
are	269	538	283	547	595	794	4
using	51	550	74	559	595	794	4
recall	79	550	103	559	595	794	4
and	108	550	124	559	595	794	4
precision	129	550	168	559	595	794	4
as	172	550	182	559	595	794	4
metrics	186	550	219	559	595	794	4
of	223	550	232	559	595	794	4
evaluation.	236	550	285	559	595	794	4
The	51	562	68	571	595	794	4
formulas	71	562	108	571	595	794	4
used	111	562	130	571	595	794	4
are	134	562	147	571	595	794	4
presented	150	562	191	571	595	794	4
in	195	562	203	571	595	794	4
equations	206	562	247	571	595	794	4
7	251	562	255	571	595	794	4
and	259	562	275	571	595	794	4
8	278	562	283	571	595	794	4
5.2.	293	558	311	567	595	794	4
Controller	316	558	368	567	595	794	4
response	371	558	414	567	595	794	4
and	418	558	437	567	595	794	4
communication	440	558	516	567	595	794	4
respectively.	51	574	104	583	595	794	4
During	293	577	322	586	595	794	4
implementation	325	577	392	586	595	794	4
of	395	577	403	586	595	794	4
the	406	577	419	586	595	794	4
robot,	423	577	448	586	595	794	4
it	451	577	458	586	595	794	4
was	461	577	477	586	595	794	4
found	480	577	504	586	595	794	4
that	507	577	525	586	595	794	4
recall	106	600	131	609	595	794	4
=	134	600	142	609	595	794	4
VP	161	594	175	602	595	794	4
×	194	600	202	609	595	794	4
100%	204	600	228	609	595	794	4
V	146	607	152	616	595	794	4
P	154	607	160	616	595	794	4
+	164	607	172	616	595	794	4
FN	174	607	190	616	595	794	4
precision	99	632	140	641	595	794	4
=	143	632	151	641	595	794	4
VP	168	626	183	634	595	794	4
×	202	632	209	641	595	794	4
100%	211	632	235	641	595	794	4
V	155	639	160	648	595	794	4
P	163	639	169	648	595	794	4
+	173	639	180	648	595	794	4
FP	183	639	197	648	595	794	4
(7)	270	600	283	609	595	794	4
the	293	589	306	598	595	794	4
response	310	589	346	598	595	794	4
time	349	589	368	598	595	794	4
of	371	589	379	598	595	794	4
the	382	589	396	598	595	794	4
controller	399	589	440	598	595	794	4
varied	443	589	469	598	595	794	4
according	472	589	512	598	595	794	4
to	516	589	524	598	595	794	4
Bluetooth	293	601	337	610	595	794	4
transmission	340	601	396	610	595	794	4
speed,	399	601	426	610	595	794	4
the	430	601	444	610	595	794	4
results	447	601	476	610	595	794	4
of	479	601	488	610	595	794	4
the	491	601	505	610	595	794	4
test	508	601	525	610	595	794	4
are	293	613	306	622	595	794	4
presented	309	613	351	622	595	794	4
in	354	613	362	622	595	794	4
Table	366	613	390	622	595	794	4
3.	393	613	401	622	595	794	4
(8)	270	632	283	641	595	794	4
Table	292	636	318	644	595	794	4
3.	322	636	330	644	595	794	4
Delay	335	636	358	644	595	794	4
in	362	636	370	644	595	794	4
the	374	636	387	644	595	794	4
controller's	390	636	436	644	595	794	4
response	440	636	474	644	595	794	4
to	478	636	487	644	595	794	4
different	490	636	525	644	595	794	4
where:	66	654	94	662	595	794	4
TP:	66	665	83	674	595	794	4
true	86	665	104	674	595	794	4
positives.	107	665	147	674	595	794	4
FN:	66	677	83	686	595	794	4
false	86	677	105	686	595	794	4
negative.	108	677	147	686	595	794	4
FP:	66	689	82	698	595	794	4
false	85	689	104	698	595	794	4
positives.	108	689	148	698	595	794	4
We	66	713	80	722	595	794	4
realized	84	713	118	722	595	794	4
two	123	713	138	722	595	794	4
experiments,	143	713	199	722	595	794	4
offline	203	713	230	722	595	794	4
and	235	713	251	722	595	794	4
online.	255	713	285	722	595	794	4
For	51	725	66	734	595	794	4
the	70	725	85	734	595	794	4
online	89	725	116	734	595	794	4
test,	120	725	139	734	595	794	4
we	144	725	156	734	595	794	4
used	160	725	180	734	595	794	4
four	185	725	202	734	595	794	4
different	207	725	244	734	595	794	4
security	249	725	283	734	595	794	4
cameras	51	737	86	746	595	794	4
videos	90	737	117	746	595	794	4
obtained	120	737	158	746	595	794	4
from	162	737	182	746	595	794	4
Internet.	185	737	223	746	595	794	4
We	227	737	241	746	595	794	4
captured	244	737	283	746	595	794	4
transmission	293	647	343	655	595	794	4
speeds.	346	647	375	655	595	794	4
Transmission	311	671	376	680	595	794	4
Speed	380	671	410	680	595	794	4
(bauds)	414	671	452	680	595	794	4
Delay(s)	464	671	506	680	595	794	4
9600	372	688	392	697	595	794	4
14400	369	700	394	709	595	794	4
19200	369	712	394	721	595	794	4
56000	369	724	394	733	595	794	4
115200	367	736	397	745	595	794	4
≈	473	688	481	697	595	794	4
2,5	484	688	497	697	595	794	4
≈	477	700	485	709	595	794	4
2	488	700	493	709	595	794	4
≈	473	712	481	721	595	794	4
1,5	484	712	497	721	595	794	4
1	482	724	487	733	595	794	4
≈	477	736	485	745	595	794	4
0	488	736	493	745	595	794	4
Luna	74	30	96	39	595	794	5
et	100	30	108	39	595	794	5
al.	112	30	122	39	595	794	5
/	124	30	129	39	595	794	5
Mobile	133	30	161	39	595	794	5
robot	165	30	186	39	595	794	5
with	190	30	208	39	595	794	5
vision	211	30	237	39	595	794	5
based	240	30	263	39	595	794	5
navigation	266	30	311	39	595	794	5
and	315	30	331	39	595	794	5
pedestrian	334	30	378	39	595	794	5
detection	382	30	420	39	595	794	5
It	86	67	93	76	595	794	5
should	97	67	125	76	595	794	5
be	128	67	139	76	595	794	5
noted	142	67	166	76	595	794	5
that	170	67	188	76	595	794	5
the	191	67	205	76	595	794	5
delays	208	67	235	76	595	794	5
were	238	67	258	76	595	794	5
measured	261	67	303	76	595	794	5
according	71	79	113	88	595	794	5
to	116	79	125	88	595	794	5
the	129	79	143	88	595	794	5
perceived	146	79	187	88	595	794	5
response	190	79	228	88	595	794	5
of	231	79	239	88	595	794	5
the	243	79	257	88	595	794	5
controller.	260	79	305	88	595	794	5
A	70	91	78	100	595	794	5
transmission	81	91	135	100	595	794	5
rate	138	91	156	100	595	794	5
of	159	91	167	100	595	794	5
115200	170	91	200	100	595	794	5
baud	203	91	224	100	595	794	5
was	228	91	243	100	595	794	5
chosen	247	91	275	100	595	794	5
based	278	91	303	100	595	794	5
on	71	103	81	112	595	794	5
results.	85	103	116	112	595	794	5
6.	71	130	81	140	595	794	5
Conclusiones	88	130	164	140	595	794	5
71	533	29	544	39	595	794	5
[5]	318	67	328	76	595	794	5
U.	333	67	343	76	595	794	5
Franke,	349	67	381	76	595	794	5
D.	387	67	397	76	595	794	5
Gavrila,	402	67	438	76	595	794	5
S.	443	67	451	76	595	794	5
Görzig,	456	67	489	76	595	794	5
F.	494	67	503	76	595	794	5
Lindner,	508	67	546	76	595	794	5
F.	333	79	342	88	595	794	5
Paetzold,	345	79	384	88	595	794	5
and	387	79	402	88	595	794	5
C.	405	79	415	88	595	794	5
Wöhler,	417	79	451	88	595	794	5
“Autonomous	453	79	512	88	595	794	5
driving	514	79	544	88	595	794	5
goes	333	91	352	100	595	794	5
downtown,”	355	91	408	100	595	794	5
IEEE	412	91	436	100	595	794	5
Intelligent	440	91	485	100	595	794	5
Systems	489	91	524	100	595	794	5
and	528	91	544	100	595	794	5
Their	331	103	355	112	595	794	5
Applications,	358	103	413	112	595	794	5
vol.	416	103	431	112	595	794	5
13,	435	103	447	112	595	794	5
no.	450	103	463	112	595	794	5
6,	466	103	474	112	595	794	5
pp.	477	103	491	112	595	794	5
40–48,	494	103	521	112	595	794	5
1998.	524	103	546	112	595	794	5
[6]	318	125	328	133	595	794	5
U.	333	125	343	133	595	794	5
Franke	347	125	378	133	595	794	5
and	382	125	398	133	595	794	5
I.	402	125	408	133	595	794	5
Kutzbach,	412	125	458	133	595	794	5
“Fast	462	125	485	133	595	794	5
stereo	489	125	515	133	595	794	5
based	519	125	544	133	595	794	5
object	333	137	359	145	595	794	5
detection	362	137	402	145	595	794	5
for	405	137	417	145	595	794	5
stopgo	420	137	455	145	595	794	5
traffic,”	459	137	492	145	595	794	5
IEEE	495	137	519	145	595	794	5
Intel-	522	137	545	145	595	794	5
ligent	332	148	356	157	595	794	5
Vehicles	360	148	395	157	595	794	5
Symposium,	398	148	450	157	595	794	5
pp.	453	148	467	157	595	794	5
339–344,	471	148	508	157	595	794	5
1996.	511	148	534	157	595	794	5
The	71	153	88	162	595	794	5
HOG	91	153	115	162	595	794	5
algorithm	118	153	161	162	595	794	5
has	164	153	179	162	595	794	5
better	182	153	209	162	595	794	5
performance	213	153	267	162	595	794	5
off-line;	271	153	304	162	595	794	5
however,	71	165	108	173	595	794	5
the	111	165	125	173	595	794	5
implementation	128	165	195	173	595	794	5
represents	199	165	242	173	595	794	5
a	245	165	250	173	595	794	5
higher	253	165	280	173	595	794	5
com-	283	165	304	173	595	794	5
[7]	318	170	328	179	595	794	5
W.	333	170	346	179	595	794	5
Aguilar	351	170	384	179	595	794	5
and	389	170	405	179	595	794	5
S.	410	170	418	179	595	794	5
Morales,	423	170	460	179	595	794	5
“3D	465	170	482	179	595	794	5
Environment	487	170	545	179	595	794	5
putational	71	177	116	185	595	794	5
cost.	119	177	139	185	595	794	5
LBP	142	177	162	185	595	794	5
had	166	177	182	185	595	794	5
a	185	177	190	185	595	794	5
lower	193	177	216	185	595	794	5
performance	219	177	273	185	595	794	5
offline	276	177	303	185	595	794	5
Mapping	333	182	371	191	595	794	5
Using	374	182	398	191	595	794	5
the	401	182	414	191	595	794	5
Kinect	417	182	446	191	595	794	5
V2	449	182	461	191	595	794	5
and	464	182	480	191	595	794	5
Path	483	182	503	191	595	794	5
Planning	506	182	544	191	595	794	5
but	71	189	86	197	595	794	5
was	89	189	104	197	595	794	5
not	108	189	122	197	595	794	5
affected	125	189	159	197	595	794	5
for	162	189	174	197	595	794	5
online	177	189	203	197	595	794	5
experiments.	206	189	260	197	595	794	5
The	264	189	280	197	595	794	5
false	284	189	303	197	595	794	5
Based	333	194	359	203	595	794	5
on	363	194	374	203	595	794	5
RRT	378	194	399	203	595	794	5
Algorithms,”	403	194	460	203	595	794	5
Electronics,	463	194	515	203	595	794	5
vol.	518	194	534	203	595	794	5
5,	538	194	546	203	595	794	5
positive	71	201	104	209	595	794	5
detection	108	201	148	209	595	794	5
rate	151	201	168	209	595	794	5
involves	171	201	206	209	595	794	5
low	209	201	223	209	595	794	5
precision.	227	201	268	209	595	794	5
no.	333	206	346	215	595	794	5
4,	350	206	357	215	595	794	5
p.	361	206	369	215	595	794	5
70,	372	206	385	215	595	794	5
2016.	388	206	411	215	595	794	5
Classical	86	212	124	221	595	794	5
method	127	212	160	221	595	794	5
of	163	212	171	221	595	794	5
PI	174	212	185	221	595	794	5
control	188	212	218	221	595	794	5
provided	222	212	259	221	595	794	5
a	262	212	267	221	595	794	5
smooth	271	212	303	221	595	794	5
braking	71	224	104	233	595	794	5
control	108	224	139	233	595	794	5
to	143	224	152	233	595	794	5
keep	156	224	176	233	595	794	5
a	179	224	184	233	595	794	5
safe	188	224	205	233	595	794	5
distance	209	224	245	233	595	794	5
between	249	224	285	233	595	794	5
the	288	224	303	233	595	794	5
[8]	318	228	328	237	595	794	5
W.	333	228	346	237	595	794	5
G.	349	228	360	237	595	794	5
Aguilar	363	228	395	237	595	794	5
and	398	228	414	237	595	794	5
C.	417	228	427	237	595	794	5
Angulo,	430	228	464	237	595	794	5
“Estabilización	467	228	531	237	595	794	5
de	534	228	544	237	595	794	5
mobile	71	236	100	245	595	794	5
robot	104	236	128	245	595	794	5
and	131	236	148	245	595	794	5
the	151	236	165	245	595	794	5
pedestrian.	169	236	218	245	595	794	5
On-Off	222	236	253	245	595	794	5
control	256	236	287	245	595	794	5
for	291	236	303	245	595	794	5
vídeo	333	240	355	249	595	794	5
en	358	240	368	249	595	794	5
micro	371	240	395	249	595	794	5
vehículos	398	240	437	249	595	794	5
aéreos	440	240	466	249	595	794	5
y	469	240	474	249	595	794	5
su	477	240	486	249	595	794	5
aplicación	489	240	531	249	595	794	5
en	534	240	544	249	595	794	5
the	71	248	85	257	595	794	5
horizontal	88	248	131	257	595	794	5
position	134	248	169	257	595	794	5
is	172	248	179	257	595	794	5
a	182	248	187	257	595	794	5
simple	190	248	218	257	595	794	5
controller	221	248	262	257	595	794	5
that	265	248	283	257	595	794	5
had	287	248	303	257	595	794	5
la	333	252	341	261	595	794	5
detección	344	252	384	261	595	794	5
de	388	252	398	261	595	794	5
caras,”	401	252	431	261	595	794	5
pp.	434	252	448	261	595	794	5
155–160,	451	252	489	261	595	794	5
2014.	492	252	515	261	595	794	5
a	71	260	76	269	595	794	5
good	79	260	101	269	595	794	5
performance	104	260	158	269	595	794	5
maintaining	162	260	214	269	595	794	5
the	218	260	232	269	595	794	5
direction	235	260	274	269	595	794	5
of	277	260	285	269	595	794	5
the	289	260	303	269	595	794	5
navigation.	71	272	119	281	595	794	5
[9]	318	274	328	282	595	794	5
D.	333	274	344	282	595	794	5
Gavrila,	348	274	383	282	595	794	5
J.	387	274	395	282	595	794	5
Giebel,	400	274	431	282	595	794	5
and	435	274	451	282	595	794	5
S.	456	274	464	282	595	794	5
Munder,	468	274	506	282	595	794	5
“Vision-	510	274	546	282	595	794	5
A	86	284	93	293	595	794	5
slow	96	284	114	293	595	794	5
controller	117	284	158	293	595	794	5
response	160	284	197	293	595	794	5
endangered	200	284	248	293	595	794	5
the	251	284	264	293	595	794	5
safety	267	284	292	293	595	794	5
of	295	284	303	293	595	794	5
based	333	286	358	294	595	794	5
pedestrian	362	286	408	294	595	794	5
detection:	412	286	456	294	595	794	5
the	460	286	474	294	595	794	5
PROTECTOR	478	286	544	294	595	794	5
pedestrians,	71	296	123	305	595	794	5
so	126	296	135	305	595	794	5
it	138	296	145	305	595	794	5
was	148	296	164	305	595	794	5
necessary	167	296	208	305	595	794	5
to	212	296	220	305	595	794	5
increase	224	296	258	305	595	794	5
the	262	296	275	305	595	794	5
trans-	279	296	304	305	595	794	5
system,”	333	298	371	306	595	794	5
IEEE	375	298	399	306	595	794	5
Intelligent	403	298	449	306	595	794	5
Vehicles	453	298	489	306	595	794	5
Symposium,	493	298	546	306	595	794	5
mission	71	308	104	317	595	794	5
speed	107	308	131	317	595	794	5
obtaining	135	308	176	317	595	794	5
better	179	308	206	317	595	794	5
results.	210	308	241	317	595	794	5
The	244	308	262	317	595	794	5
response	265	308	303	317	595	794	5
2004,	332	310	355	318	595	794	5
pp.	359	310	373	318	595	794	5
13–18,	376	310	404	318	595	794	5
2004.	407	310	430	318	595	794	5
time	71	320	91	329	595	794	5
also	95	320	112	329	595	794	5
was	116	320	132	329	595	794	5
affected	136	320	171	329	595	794	5
by	175	320	186	329	595	794	5
the	190	320	204	329	595	794	5
parameters	208	320	257	329	595	794	5
of	261	320	270	329	595	794	5
the	274	320	288	329	595	794	5
PI	292	320	303	329	595	794	5
controller.	71	332	114	341	595	794	5
People	117	332	145	341	595	794	5
tracking	149	332	184	341	595	794	5
algorithms	187	332	232	341	595	794	5
have	235	332	255	341	595	794	5
benefits	258	332	291	341	595	794	5
in	294	332	303	341	595	794	5
[10]	313	331	328	340	595	794	5
a.	333	331	341	340	595	794	5
Shashua,	344	331	382	340	595	794	5
Y.	385	331	396	340	595	794	5
Gdalyahu,	399	331	443	340	595	794	5
and	446	331	462	340	595	794	5
G.	465	331	476	340	595	794	5
Hayun,	479	331	510	340	595	794	5
“Pedes-	513	331	546	340	595	794	5
trian	333	343	355	352	595	794	5
detection	360	343	400	352	595	794	5
for	406	343	418	352	595	794	5
driving	423	343	454	352	595	794	5
assistance	460	343	503	352	595	794	5
systems:	508	343	546	352	595	794	5
several	71	344	101	353	595	794	5
applications.	104	344	160	353	595	794	5
We	163	344	178	353	595	794	5
can	181	344	196	353	595	794	5
improve	200	344	235	353	595	794	5
the	238	344	252	353	595	794	5
algorithms	256	344	303	353	595	794	5
single-frame	333	355	386	364	595	794	5
classification	390	355	447	364	595	794	5
and	450	355	467	364	595	794	5
system	470	355	501	364	595	794	5
level	505	355	524	364	595	794	5
per-	528	355	546	364	595	794	5
for	71	356	83	365	595	794	5
pedestrian	87	356	133	365	595	794	5
detection,	138	356	181	365	595	794	5
based	185	356	210	365	595	794	5
on	215	356	225	365	595	794	5
obtained	230	356	268	365	595	794	5
results,	272	356	304	365	595	794	5
formance,”	333	367	379	376	595	794	5
IEEE	382	367	406	376	595	794	5
Intelligent	410	367	453	376	595	794	5
Vehicles	457	367	491	376	595	794	5
Symposium,	495	367	546	376	595	794	5
with	71	368	89	377	595	794	5
some	93	368	114	377	595	794	5
techniques	117	368	162	377	595	794	5
like	165	368	180	377	595	794	5
specialized	183	368	228	377	595	794	5
training	231	368	265	377	595	794	5
or	268	368	277	377	595	794	5
video	280	368	303	377	595	794	5
2004,	332	379	355	388	595	794	5
pp.	359	379	373	388	595	794	5
1–6,	376	379	394	388	595	794	5
2004.	397	379	420	388	595	794	5
treating,	71	380	107	389	595	794	5
but	111	380	125	389	595	794	5
this	129	380	145	389	595	794	5
involves	148	380	181	389	595	794	5
an	185	380	195	389	595	794	5
increased	198	380	238	389	595	794	5
computational	241	380	303	389	595	794	5
cost.	71	392	91	401	595	794	5
[11]	313	401	328	410	595	794	5
D.	333	401	344	410	595	794	5
Gerónimo,	348	401	394	410	595	794	5
A.	399	401	409	410	595	794	5
M.	414	401	426	410	595	794	5
López,	430	401	459	410	595	794	5
A.	464	401	474	410	595	794	5
D.	479	401	489	410	595	794	5
Sappa,	494	401	523	410	595	794	5
and	528	401	544	410	595	794	5
T.	333	413	343	422	595	794	5
Graf,	346	413	369	422	595	794	5
“Survey	373	413	408	422	595	794	5
of	411	413	419	422	595	794	5
pedestrian	423	413	469	422	595	794	5
detection	472	413	513	422	595	794	5
for	516	413	528	422	595	794	5
ad-	532	413	546	422	595	794	5
Acknowledgments	71	419	176	429	595	794	5
vanced	333	425	363	434	595	794	5
driver	366	425	392	434	595	794	5
assistance	396	425	439	434	595	794	5
systems,”	443	425	485	434	595	794	5
IEEE	488	425	513	434	595	794	5
Trans-	516	425	545	434	595	794	5
actions	332	437	363	446	595	794	5
on	366	437	377	446	595	794	5
Pattern	381	437	413	446	595	794	5
Analysis	417	437	453	446	595	794	5
and	456	437	472	446	595	794	5
Machine	475	437	512	446	595	794	5
Intelli-	516	437	545	446	595	794	5
This	71	442	90	451	595	794	5
work	94	442	116	451	595	794	5
is	120	442	127	451	595	794	5
part	131	442	150	451	595	794	5
of	154	442	162	451	595	794	5
the	166	442	180	451	595	794	5
research	185	442	221	451	595	794	5
project	225	442	256	451	595	794	5
“Vehículo	260	442	303	451	595	794	5
gence,	332	449	359	458	595	794	5
vol.	362	449	377	458	595	794	5
32,	381	449	393	458	595	794	5
no.	397	449	410	458	595	794	5
7,	413	449	421	458	595	794	5
pp.	424	449	438	458	595	794	5
1239–1258,	442	449	489	458	595	794	5
2010.	492	449	515	458	595	794	5
terrestre	71	454	107	463	595	794	5
multipropósito	111	454	174	463	595	794	5
no	178	454	188	463	595	794	5
tripulado	191	454	231	463	595	794	5
(MultiNavCar)”	235	454	304	463	595	794	5
from	71	466	92	475	595	794	5
Universidad	95	466	148	475	595	794	5
de	151	466	161	475	595	794	5
las	165	466	176	475	595	794	5
Fuerzas	180	466	213	475	595	794	5
Armadas	217	466	257	475	595	794	5
ESPE,	260	466	289	475	595	794	5
di-	292	466	304	475	595	794	5
[12]	313	471	328	479	595	794	5
W.	333	471	346	479	595	794	5
G.	349	471	359	479	595	794	5
Aguilar	362	471	394	479	595	794	5
and	397	471	412	479	595	794	5
C.	415	471	425	479	595	794	5
Angulo,	428	471	461	479	595	794	5
“Real-Time	464	471	513	479	595	794	5
Model-	516	471	546	479	595	794	5
rected	71	478	97	486	595	794	5
by	101	478	111	486	595	794	5
Dr.	115	478	129	486	595	794	5
Wilbert	132	478	166	486	595	794	5
G.	169	478	180	486	595	794	5
Aguilar.	183	478	218	486	595	794	5
Based	333	483	359	491	595	794	5
Video	363	483	389	491	595	794	5
Stabilization	393	483	449	491	595	794	5
for	453	483	465	491	595	794	5
Microaerial	469	483	519	491	595	794	5
Vehi-	523	483	546	491	595	794	5
cles,”	333	495	356	503	595	794	5
Neural	359	495	387	503	595	794	5
Processing	390	495	434	503	595	794	5
Letters,	437	495	469	503	595	794	5
vol.	472	495	487	503	595	794	5
43,	490	495	503	503	595	794	5
no.	506	495	519	503	595	794	5
2,	522	495	530	503	595	794	5
pp.	533	495	546	503	595	794	5
References	71	505	134	515	595	794	5
459–477,	333	507	370	515	595	794	5
2016.	374	507	396	515	595	794	5
[1]	76	528	86	536	595	794	5
M.	91	528	103	536	595	794	5
Bertozzi,	107	528	145	536	595	794	5
A.	149	528	159	536	595	794	5
Broggi,	162	528	194	536	595	794	5
and	197	528	213	536	595	794	5
A.	217	528	227	536	595	794	5
Fascioli,	230	528	265	536	595	794	5
“Vision-	269	528	304	536	595	794	5
[13]	313	528	328	537	595	794	5
——,	333	528	356	537	595	794	5
“Real-time	363	528	411	537	595	794	5
video	418	528	442	537	595	794	5
stabilization	449	528	503	537	595	794	5
without	510	528	545	537	595	794	5
based	91	540	116	548	595	794	5
intelligent	121	540	165	548	595	794	5
vehicles:	169	540	206	548	595	794	5
State	210	540	233	548	595	794	5
of	238	540	246	548	595	794	5
the	250	540	264	548	595	794	5
art	269	540	282	548	595	794	5
and	286	540	303	548	595	794	5
phantom	333	540	372	549	595	794	5
movements	377	540	426	549	595	794	5
for	430	540	442	549	595	794	5
micro	447	540	472	549	595	794	5
aerial	476	540	500	549	595	794	5
vehicles,”	505	540	546	549	595	794	5
perspectives,”	91	552	150	560	595	794	5
Robotics	153	552	189	560	595	794	5
and	192	552	208	560	595	794	5
Autonomous	211	552	264	560	595	794	5
Systems,	267	552	304	560	595	794	5
EURASIP	332	552	378	561	595	794	5
Journal	381	552	415	561	595	794	5
on	418	552	429	561	595	794	5
Image	432	552	459	561	595	794	5
and	462	552	478	561	595	794	5
Video	482	552	506	561	595	794	5
Process-	510	552	545	561	595	794	5
vol.	91	564	107	572	595	794	5
32,	110	564	123	572	595	794	5
no.	126	564	139	572	595	794	5
1,	143	564	150	572	595	794	5
pp.	154	564	167	572	595	794	5
1–16,	171	564	193	572	595	794	5
2000.	197	564	219	572	595	794	5
ing,	332	564	348	573	595	794	5
vol.	352	564	367	573	595	794	5
2014,	371	564	393	573	595	794	5
no.	397	564	410	573	595	794	5
1,	413	564	421	573	595	794	5
p.	424	564	433	573	595	794	5
46,	436	564	449	573	595	794	5
2014.	452	564	475	573	595	794	5
[2]	76	582	86	590	595	794	5
W.	91	582	104	590	595	794	5
Aguilar	107	582	139	590	595	794	5
and	142	582	157	590	595	794	5
C.	160	582	170	590	595	794	5
Angulo,	173	582	206	590	595	794	5
“Compensación	209	582	276	590	595	794	5
de	278	582	288	590	595	794	5
los	291	582	303	590	595	794	5
[14]	313	586	328	595	595	794	5
——,	333	586	356	595	595	794	5
“Robust	359	586	396	595	595	794	5
Video	399	586	424	595	595	794	5
Stabilization	428	586	483	595	595	794	5
based	486	586	511	595	595	794	5
on	514	586	525	595	595	794	5
Mo-	528	586	546	595	595	794	5
Efectos	91	594	122	602	595	794	5
Generados	125	594	170	602	595	794	5
en	173	594	183	602	595	794	5
la	185	594	193	602	595	794	5
Imagen	196	594	227	602	595	794	5
por	230	594	245	602	595	794	5
el	248	594	255	602	595	794	5
Control	258	594	290	602	595	794	5
de	293	594	303	602	595	794	5
tion	333	598	350	607	595	794	5
Intention	353	598	392	607	595	794	5
for	395	598	406	607	595	794	5
Micro	409	598	434	607	595	794	5
Aerial	437	598	463	607	595	794	5
Vehicles,”	466	598	507	607	595	794	5
Systems	510	598	544	607	595	794	5
Navegación	91	605	142	614	595	794	5
del	145	605	158	614	595	794	5
Robot	162	605	190	614	595	794	5
Aibo	193	605	215	614	595	794	5
ERS	218	605	238	614	595	794	5
7,”	242	605	255	614	595	794	5
Memorias	259	605	303	614	595	794	5
Signals	333	610	364	619	595	794	5
and	368	610	384	619	595	794	5
Devices	388	610	422	619	595	794	5
(SSD),	426	610	456	619	595	794	5
2014	461	610	482	619	595	794	5
International	486	610	544	619	595	794	5
del	91	617	103	626	595	794	5
VII	106	617	122	626	595	794	5
Congreso	125	617	165	626	595	794	5
de	169	617	178	626	595	794	5
Ciencia	182	617	215	626	595	794	5
y	219	617	223	626	595	794	5
Tecnología	227	617	273	626	595	794	5
ESPE	276	617	302	626	595	794	5
Multi-Conference	332	622	407	631	595	794	5
on,	411	622	424	631	595	794	5
pp.	428	622	441	631	595	794	5
1	445	622	450	631	595	794	5
–	453	622	458	631	595	794	5
6,	461	622	469	631	595	794	5
2014.	472	622	495	631	595	794	5
2012,	91	629	114	638	595	794	5
no.	117	629	130	638	595	794	5
JUNE,	134	629	163	638	595	794	5
pp.	167	629	181	638	595	794	5
165–170,	184	629	221	638	595	794	5
2012.	225	629	247	638	595	794	5
[3]	76	647	86	656	595	794	5
——,	91	647	114	656	595	794	5
“Compensación	118	647	187	656	595	794	5
y	191	647	196	656	595	794	5
Aprendizaje	200	647	253	656	595	794	5
de	257	647	267	656	595	794	5
Efectos	270	647	303	656	595	794	5
[15]	313	644	328	652	595	794	5
T.	333	644	343	652	595	794	5
Teixeira,	348	644	386	652	595	794	5
D.	391	644	401	652	595	794	5
Jung,	406	644	430	652	595	794	5
and	435	644	451	652	595	794	5
A.	456	644	467	652	595	794	5
Savvides,	471	644	512	652	595	794	5
“Task-	517	644	546	652	595	794	5
ing	333	656	346	664	595	794	5
networked	349	656	392	664	595	794	5
cctv	395	656	413	664	595	794	5
cameras	416	656	450	664	595	794	5
and	453	656	469	664	595	794	5
mobile	472	656	500	664	595	794	5
phones	503	656	533	664	595	794	5
to	536	656	544	664	595	794	5
Generados	91	659	138	668	595	794	5
en	143	659	153	668	595	794	5
la	158	659	166	668	595	794	5
Imagen	172	659	204	668	595	794	5
durante	209	659	244	668	595	794	5
el	249	659	256	668	595	794	5
Desplaza-	261	659	304	668	595	794	5
identify	333	668	365	676	595	794	5
and	368	668	384	676	595	794	5
localize	387	668	418	676	595	794	5
multiple	420	668	455	676	595	794	5
people,”	458	668	493	676	595	794	5
Proceedings	496	668	544	676	595	794	5
miento	91	671	120	680	595	794	5
de	123	671	133	680	595	794	5
un	135	671	146	680	595	794	5
Robot,”	148	671	182	680	595	794	5
X	185	671	192	680	595	794	5
Simposio	195	671	233	680	595	794	5
CEA	236	671	256	680	595	794	5
Ingeniería	259	671	303	680	595	794	5
of	332	680	340	688	595	794	5
the	344	680	357	688	595	794	5
12th	360	680	379	688	595	794	5
ACM	383	680	406	688	595	794	5
.	409	680	412	688	595	794	5
.	414	680	417	688	595	794	5
.	418	680	421	688	595	794	5
,	423	680	426	688	595	794	5
pp.	429	680	443	688	595	794	5
213–222,	446	680	484	688	595	794	5
2010.	487	680	510	688	595	794	5
de	91	683	100	692	595	794	5
Control	104	683	136	692	595	794	5
1,2	140	683	153	692	595	794	5
Marzo,	157	683	187	692	595	794	5
2012,	191	683	214	692	595	794	5
pp.	217	683	231	692	595	794	5
165–170,	234	683	272	692	595	794	5
2012.	275	683	298	692	595	794	5
[4]	76	701	86	710	595	794	5
A.	91	701	102	710	595	794	5
Geiger,	107	701	138	710	595	794	5
P.	143	701	152	710	595	794	5
Lenz,	157	701	181	710	595	794	5
and	185	701	202	710	595	794	5
R.	207	701	217	710	595	794	5
Urtasun,	222	701	260	710	595	794	5
“Are	265	701	286	710	595	794	5
we	291	701	303	710	595	794	5
[16]	313	701	328	710	595	794	5
H.	333	701	343	710	595	794	5
Torresan,	349	701	391	710	595	794	5
“Advanced	397	701	445	710	595	794	5
surveillance	451	701	503	710	595	794	5
systems:	508	701	546	710	595	794	5
ready	91	713	116	722	595	794	5
for	119	713	132	722	595	794	5
Autonomous	135	713	191	722	595	794	5
Driving?	194	713	233	722	595	794	5
The	236	713	254	722	595	794	5
KITTI	257	713	287	722	595	794	5
Vi-	290	713	304	722	595	794	5
combining	333	713	378	722	595	794	5
video	381	713	404	722	595	794	5
and	408	713	424	722	595	794	5
thermal	427	713	461	722	595	794	5
imagery	465	713	500	722	595	794	5
for	503	713	515	722	595	794	5
pedes-	518	713	546	722	595	794	5
sion	91	725	109	734	595	794	5
Benchmark	114	725	164	734	595	794	5
Suite,”	169	725	199	734	595	794	5
Computer	204	725	248	734	595	794	5
Vision	253	725	281	734	595	794	5
and	286	725	302	734	595	794	5
trian	333	725	354	734	595	794	5
detection,”	358	725	405	734	595	794	5
Proceedings	409	725	459	734	595	794	5
of	463	725	471	734	595	794	5
SPIE,	474	725	500	734	595	794	5
vol.	504	725	519	734	595	794	5
5405,	523	725	546	734	595	794	5
Pattern	91	737	124	746	595	794	5
Recognition,	127	737	180	746	595	794	5
pp.	183	737	197	746	595	794	5
3354–3361,	200	737	248	746	595	794	5
2012.	251	737	274	746	595	794	5
pp.	333	737	347	746	595	794	5
506–515,	350	737	388	746	595	794	5
2004.	391	737	414	746	595	794	5
72	51	29	62	39	595	794	6
INGENIUS	337	30	396	39	595	794	6
N.	399	30	409	39	595	794	6
◦	409	28	413	35	595	794	6
17,	417	30	430	39	595	794	6
Enero-Junio	433	30	486	39	595	794	6
de	491	30	501	39	595	794	6
2017	504	30	524	39	595	794	6
[17]	51	67	67	76	595	794	6
L.	72	67	80	76	595	794	6
Zhang,	83	67	112	76	595	794	6
B.	115	67	125	76	595	794	6
Wu,	128	67	145	76	595	794	6
and	148	67	163	76	595	794	6
R.	166	67	176	76	595	794	6
Nevatia,	179	67	214	76	595	794	6
“Pedestrian	217	67	267	76	595	794	6
De-	269	67	284	76	595	794	6
classification,”	313	67	376	76	595	794	6
Advances	380	67	420	76	595	794	6
in	424	67	432	76	595	794	6
Pattern	436	67	469	76	595	794	6
Recognition,	473	67	526	76	595	794	6
tection	72	79	101	88	595	794	6
in	105	79	113	88	595	794	6
Infrared	116	79	151	88	595	794	6
Images	154	79	185	88	595	794	6
based	188	79	212	88	595	794	6
on	216	79	226	88	595	794	6
Local	230	79	253	88	595	794	6
Shape	256	79	283	88	595	794	6
vol.	313	79	328	88	595	794	6
2013,	332	79	354	88	595	794	6
pp.	358	79	372	88	595	794	6
399–408,	375	79	413	88	595	794	6
2001.	416	79	439	88	595	794	6
Features,”	72	91	116	100	595	794	6
2007	120	91	141	100	595	794	6
IEEE	144	91	169	100	595	794	6
Conference	172	91	221	100	595	794	6
on	225	91	235	100	595	794	6
Computer	239	91	283	100	595	794	6
[27]	293	99	308	108	595	794	6
C.	313	99	323	108	595	794	6
Wojek	326	99	353	108	595	794	6
and	356	99	372	108	595	794	6
B.	375	99	384	108	595	794	6
Schiele,	387	99	419	108	595	794	6
“A	422	99	434	108	595	794	6
Performance	437	99	491	108	595	794	6
Evalua-	494	99	526	108	595	794	6
Vision	70	103	97	112	595	794	6
and	101	103	117	112	595	794	6
Pattern	120	103	153	112	595	794	6
Recognition,	157	103	209	112	595	794	6
pp.	213	103	227	112	595	794	6
0–7,	230	103	248	112	595	794	6
2007.	251	103	274	112	595	794	6
tion	313	111	330	120	595	794	6
of	332	111	340	120	595	794	6
Single	343	111	368	120	595	794	6
and	370	111	386	120	595	794	6
Multi-feature	389	111	445	120	595	794	6
People	447	111	476	120	595	794	6
Detection,”	478	111	526	120	595	794	6
[18]	51	121	67	130	595	794	6
D.	72	121	82	130	595	794	6
M.	84	121	96	130	595	794	6
Gavrila,	99	121	133	130	595	794	6
“Pedestrian	136	121	186	130	595	794	6
Detection	188	121	229	130	595	794	6
from	232	121	252	130	595	794	6
a	255	121	260	130	595	794	6
Mov-	262	121	284	130	595	794	6
Pattern	313	123	346	131	595	794	6
Recognition,	351	123	405	131	595	794	6
vol.	409	123	425	131	595	794	6
11,	430	123	443	131	595	794	6
no.	448	123	461	131	595	794	6
2,	466	123	474	131	595	794	6
pp.	479	123	493	131	595	794	6
82–91,	498	123	526	131	595	794	6
ing	72	133	85	142	595	794	6
Vehicle,”	88	133	127	142	595	794	6
System,	130	133	163	142	595	794	6
pp.	167	133	180	142	595	794	6
37–49,	184	133	211	142	595	794	6
2000.	215	133	237	142	595	794	6
2008.	313	135	336	143	595	794	6
[19]	51	152	67	161	595	794	6
M.	72	152	84	161	595	794	6
Enzweiler	89	152	131	161	595	794	6
and	136	152	153	161	595	794	6
D.	158	152	168	161	595	794	6
M.	173	152	185	161	595	794	6
Gavrila,	190	152	226	161	595	794	6
“Monocular	231	152	283	161	595	794	6
[28]	293	155	308	163	595	794	6
X.	313	155	323	163	595	794	6
Wang,	326	155	353	163	595	794	6
T.	356	155	366	163	595	794	6
X.	369	155	379	163	595	794	6
Han,	382	155	403	163	595	794	6
and	406	155	422	163	595	794	6
S.	425	155	433	163	595	794	6
Yan,	436	155	455	163	595	794	6
“An	458	155	476	163	595	794	6
HOG-LBP	479	155	524	163	595	794	6
pedestrian	72	164	118	173	595	794	6
detection:	122	164	165	173	595	794	6
Survey	169	164	199	173	595	794	6
and	203	164	220	173	595	794	6
experiments,”	224	164	285	173	595	794	6
human	313	167	343	175	595	794	6
detector	346	167	381	175	595	794	6
with	384	167	404	175	595	794	6
partial	407	167	436	175	595	794	6
occlusion	439	167	478	175	595	794	6
handling,”	481	167	526	175	595	794	6
IEEE	71	176	95	185	595	794	6
Transactions	98	176	154	185	595	794	6
on	157	176	168	185	595	794	6
Pattern	171	176	204	185	595	794	6
Analysis	208	176	244	185	595	794	6
and	247	176	263	185	595	794	6
Ma-	266	176	284	185	595	794	6
Computer	312	178	355	187	595	794	6
Vision,	359	178	390	187	595	794	6
2009	393	178	414	187	595	794	6
IEEE	417	178	441	187	595	794	6
12th	445	178	464	187	595	794	6
International	467	178	524	187	595	794	6
chine	71	188	93	197	595	794	6
Intelligence,	97	188	149	197	595	794	6
vol.	153	188	168	197	595	794	6
31,	171	188	184	197	595	794	6
no.	187	188	200	197	595	794	6
12,	204	188	216	197	595	794	6
pp.	220	188	233	197	595	794	6
2179–2195,	237	188	284	197	595	794	6
Conference	312	190	360	199	595	794	6
on,	364	190	377	199	595	794	6
no.	380	190	394	199	595	794	6
Iccv,	397	190	417	199	595	794	6
pp.	421	190	435	199	595	794	6
32–39,	438	190	466	199	595	794	6
2009.	469	190	492	199	595	794	6
2009.	71	200	94	209	595	794	6
[29]	293	210	308	219	595	794	6
P.	313	210	322	219	595	794	6
Dollar,	326	210	356	219	595	794	6
C.	360	210	370	219	595	794	6
Wojek,	374	210	405	219	595	794	6
B.	409	210	419	219	595	794	6
Schiele,	422	210	455	219	595	794	6
and	459	210	476	219	595	794	6
P.	479	210	488	219	595	794	6
Perona,	492	210	526	219	595	794	6
[20]	51	218	67	227	595	794	6
O.	72	218	82	227	595	794	6
H.	85	218	95	227	595	794	6
Jafari,	98	218	125	227	595	794	6
D.	128	218	138	227	595	794	6
Mitzel,	141	218	170	227	595	794	6
and	173	218	189	227	595	794	6
B.	192	218	202	227	595	794	6
Leibe,	205	218	231	227	595	794	6
“Real-Time	234	218	283	227	595	794	6
“Pedestrian	312	222	363	231	595	794	6
detection:	366	222	409	231	595	794	6
an	412	222	423	231	595	794	6
evaluation	426	222	471	231	595	794	6
of	474	222	483	231	595	794	6
the	486	222	500	231	595	794	6
state	503	222	524	231	595	794	6
RGB-D	72	230	104	239	595	794	6
based	107	230	131	239	595	794	6
People	135	230	163	239	595	794	6
Detection	166	230	208	239	595	794	6
and	211	230	227	239	595	794	6
Tracking	230	230	268	239	595	794	6
for	271	230	283	239	595	794	6
of	313	234	321	243	595	794	6
the	326	234	340	243	595	794	6
art,”	345	234	365	243	595	794	6
IEEE	370	234	394	243	595	794	6
Trans.	399	234	427	243	595	794	6
Pattern	432	234	466	243	595	794	6
Anal.	470	234	494	243	595	794	6
Mach.	499	234	526	243	595	794	6
Mobile	72	242	101	251	595	794	6
Robots	104	242	135	251	595	794	6
and	138	242	154	251	595	794	6
Head-Worn	158	242	207	251	595	794	6
Cameras,”	210	242	255	251	595	794	6
IEEE	258	242	282	251	595	794	6
Intell.,	313	246	342	255	595	794	6
vol.	345	246	360	255	595	794	6
34,	364	246	377	255	595	794	6
no.	380	246	393	255	595	794	6
4,	396	246	404	255	595	794	6
pp.	408	246	421	255	595	794	6
743–61,	425	246	457	255	595	794	6
2012.	461	246	483	255	595	794	6
International	71	254	127	263	595	794	6
Conference	131	254	179	263	595	794	6
on	182	254	193	263	595	794	6
Robotics	196	254	232	263	595	794	6
Automa-	247	254	284	263	595	794	6
tion	71	266	88	275	595	794	6
(ICRA),	91	266	127	275	595	794	6
no.	131	266	144	275	595	794	6
May	147	266	166	275	595	794	6
2016,	169	266	192	275	595	794	6
pp.	195	266	209	275	595	794	6
5636–5643,	212	266	259	275	595	794	6
2014.	262	266	285	275	595	794	6
[30]	293	266	308	275	595	794	6
Y.	313	266	324	275	595	794	6
Aleksandrovich,	329	266	399	275	595	794	6
“Mobile	404	266	439	275	595	794	6
Robot	444	266	471	275	595	794	6
Navigation	476	266	524	275	595	794	6
Based	313	278	339	287	595	794	6
on	342	278	353	287	595	794	6
Artificial	356	278	395	287	595	794	6
Landmarks	398	278	447	287	595	794	6
with	450	278	469	287	595	794	6
Machine	473	278	509	287	595	794	6
Vi-	513	278	526	287	595	794	6
[21]	51	285	67	294	595	794	6
M.	72	285	84	294	595	794	6
Kobilarov,	90	285	136	294	595	794	6
G.	142	285	153	294	595	794	6
Sukhatme,	159	285	206	294	595	794	6
J.	213	285	221	294	595	794	6
Hyams,	227	285	260	294	595	794	6
and	266	285	283	294	595	794	6
sion	313	290	330	299	595	794	6
System,”	333	290	371	299	595	794	6
World	374	290	400	299	595	794	6
Applied	403	290	434	299	595	794	6
Sciences	437	290	473	299	595	794	6
.	476	290	479	299	595	794	6
.	480	290	483	299	595	794	6
.	485	290	488	299	595	794	6
,	490	290	492	299	595	794	6
vol.	495	290	510	299	595	794	6
24,	513	290	526	299	595	794	6
P.	72	297	80	305	595	794	6
Batavia,	84	297	121	305	595	794	6
“People	124	297	158	305	595	794	6
tracking	161	297	197	305	595	794	6
and	201	297	217	305	595	794	6
following	220	297	260	305	595	794	6
with	263	297	283	305	595	794	6
no.	313	302	326	311	595	794	6
11,	330	302	343	311	595	794	6
pp.	346	302	360	311	595	794	6
1467–1472,	363	302	411	311	595	794	6
2013.	414	302	437	311	595	794	6
mobile	72	309	100	317	595	794	6
robot	102	309	125	317	595	794	6
using	128	309	150	317	595	794	6
an	153	309	163	317	595	794	6
omnidirectional	165	309	232	317	595	794	6
camera	234	309	265	317	595	794	6
and	267	309	283	317	595	794	6
a	72	321	77	329	595	794	6
laser,”	80	321	108	329	595	794	6
Proceedings	111	321	162	329	595	794	6
-	165	321	169	329	595	794	6
IEEE	172	321	197	329	595	794	6
International	201	321	258	329	595	794	6
Con-	262	321	284	329	595	794	6
[31]	293	322	308	331	595	794	6
H.	313	322	324	331	595	794	6
M.	334	322	346	331	595	794	6
Gross,	357	322	384	331	595	794	6
C.	395	322	405	331	595	794	6
Schroeter,	415	322	460	331	595	794	6
S.	470	322	479	331	595	794	6
Mueller,	489	322	526	331	595	794	6
ference	72	333	102	341	595	794	6
on	106	333	117	341	595	794	6
Robotics	121	333	158	341	595	794	6
and	162	333	178	341	595	794	6
Automation,	183	333	237	341	595	794	6
vol.	241	333	257	341	595	794	6
2006,	261	333	284	341	595	794	6
M.	313	334	325	343	595	794	6
Volkhardt,	329	334	377	343	595	794	6
E.	381	334	390	343	595	794	6
Einhorn,	395	334	433	343	595	794	6
A.	437	334	448	343	595	794	6
Bley,	452	334	474	343	595	794	6
C.	478	334	488	343	595	794	6
Martin,	492	334	526	343	595	794	6
no.	72	345	85	353	595	794	6
May,	88	345	109	353	595	794	6
pp.	112	345	126	353	595	794	6
557–562,	130	345	167	353	595	794	6
2006.	171	345	193	353	595	794	6
T.	313	346	323	355	595	794	6
Langner,	326	346	366	355	595	794	6
and	369	346	385	355	595	794	6
M.	389	346	401	355	595	794	6
Merten,	404	346	439	355	595	794	6
“Progress	442	346	485	355	595	794	6
in	488	346	497	355	595	794	6
devel-	500	346	526	355	595	794	6
oping	313	358	337	367	595	794	6
a	340	358	345	367	595	794	6
socially	348	358	380	367	595	794	6
assistive	383	358	418	367	595	794	6
mobile	421	358	450	367	595	794	6
home	453	358	476	367	595	794	6
robot	479	358	502	367	595	794	6
com-	505	358	526	367	595	794	6
[22]	51	363	67	372	595	794	6
Benenson,	72	363	117	372	595	794	6
Rodrigo;	120	363	158	372	595	794	6
Omran,	162	363	195	372	595	794	6
Mohamed;	199	363	245	372	595	794	6
Hosang,	249	363	284	372	595	794	6
panion	313	370	342	379	595	794	6
for	345	370	357	379	595	794	6
the	360	370	374	379	595	794	6
elderly	377	370	406	379	595	794	6
with	409	370	428	379	595	794	6
mild	431	370	450	379	595	794	6
cognitive	454	370	492	379	595	794	6
impair-	495	370	526	379	595	794	6
Jan;	71	375	89	384	595	794	6
Schiele,	93	375	125	384	595	794	6
Bernt,	128	375	155	384	595	794	6
“Ten	158	375	179	384	595	794	6
Years	182	375	206	384	595	794	6
of	209	375	217	384	595	794	6
Pedestrian	221	375	266	384	595	794	6
De-	269	375	284	384	595	794	6
ment,”	313	382	342	390	595	794	6
IEEE	345	382	369	390	595	794	6
International	372	382	428	390	595	794	6
Conference	432	382	479	390	595	794	6
on	482	382	493	390	595	794	6
Intelli-	496	382	525	390	595	794	6
tection,	72	387	104	396	595	794	6
What	107	387	131	396	595	794	6
Have	134	387	155	396	595	794	6
We	158	387	172	396	595	794	6
Learned?”	175	387	219	396	595	794	6
Proceedings	222	387	271	396	595	794	6
of	274	387	282	396	595	794	6
gent	312	394	331	402	595	794	6
Robots	334	394	363	402	595	794	6
and	366	394	382	402	595	794	6
Systems,	385	394	423	402	595	794	6
pp.	426	394	440	402	595	794	6
2430–2437,	443	394	491	402	595	794	6
2011.	494	394	517	402	595	794	6
the	71	399	84	408	595	794	6
Computer	87	399	129	408	595	794	6
Vision-ECCV	133	399	192	408	595	794	6
2014	196	399	216	408	595	794	6
Workshops,	219	399	268	408	595	794	6
pp.	271	399	285	408	595	794	6
613–627,	72	411	109	420	595	794	6
2014.	112	411	135	420	595	794	6
[32]	293	414	308	422	595	794	6
M.	313	414	325	422	595	794	6
Volkhardt,	328	414	375	422	595	794	6
S.	378	414	387	422	595	794	6
Müller,	390	414	422	422	595	794	6
C.	425	414	435	422	595	794	6
Schröter,	438	414	478	422	595	794	6
and	481	414	497	422	595	794	6
H.-M.	501	414	526	422	595	794	6
Groß,	313	426	337	434	595	794	6
Detection	340	426	380	434	595	794	6
of	383	426	391	434	595	794	6
Lounging	395	426	433	434	595	794	6
People	436	426	464	434	595	794	6
with	467	426	485	434	595	794	6
a	488	426	493	434	595	794	6
Mobile	496	426	524	434	595	794	6
[23]	51	430	67	438	595	794	6
P.	72	430	80	438	595	794	6
Viola	83	430	106	438	595	794	6
and	109	430	125	438	595	794	6
M.	128	430	140	438	595	794	6
Jones,	143	430	169	438	595	794	6
“Robust	172	430	207	438	595	794	6
real-time	210	430	248	438	595	794	6
face	252	430	268	438	595	794	6
de-	271	430	284	438	595	794	6
Robot	313	437	337	446	595	794	6
Companion.	341	437	393	446	595	794	6
Berlin,	403	437	432	446	595	794	6
Heidelberg:	435	437	485	446	595	794	6
Springer	488	437	525	446	595	794	6
tection,”	72	441	108	450	595	794	6
International	111	441	167	450	595	794	6
journal	170	441	200	450	595	794	6
of	203	441	211	450	595	794	6
computer	214	441	253	450	595	794	6
vision,	256	441	284	450	595	794	6
Berlin	313	449	340	458	595	794	6
Heidelberg,	343	449	392	458	595	794	6
2011,	396	449	418	458	595	794	6
pp.	422	449	435	458	595	794	6
328–337.	439	449	476	458	595	794	6
vol.	71	453	87	462	595	794	6
57,	90	453	103	462	595	794	6
no.	106	453	119	462	595	794	6
2,	123	453	130	462	595	794	6
pp.	134	453	148	462	595	794	6
137–154,	151	453	189	462	595	794	6
2004.	192	453	215	462	595	794	6
[24]	51	472	67	481	595	794	6
N.	72	472	82	481	595	794	6
Dalal	85	472	108	481	595	794	6
and	111	472	127	481	595	794	6
W.	130	472	143	481	595	794	6
Triggs,	146	472	175	481	595	794	6
“Histograms	178	472	231	481	595	794	6
of	234	472	242	481	595	794	6
Oriented	246	472	283	481	595	794	6
[33]	293	469	308	478	595	794	6
R.	313	469	323	478	595	794	6
E.	327	469	337	478	595	794	6
Schapire	340	469	378	478	595	794	6
and	381	469	397	478	595	794	6
Y.	401	469	411	478	595	794	6
Singer,	414	469	445	478	595	794	6
“Improved	448	469	495	478	595	794	6
boost-	498	469	526	478	595	794	6
ing	313	481	326	490	595	794	6
algorithms	328	481	374	490	595	794	6
using	376	481	398	490	595	794	6
confidence-rated	400	481	470	490	595	794	6
predictions,”	472	481	526	490	595	794	6
Gradients	72	484	115	493	595	794	6
for	120	484	132	493	595	794	6
Human	138	484	170	493	595	794	6
Detection,”	176	484	226	493	595	794	6
2005	232	484	252	493	595	794	6
IEEE	258	484	282	493	595	794	6
Machine	313	493	350	502	595	794	6
Learning,	355	493	396	502	595	794	6
vol.	401	493	417	502	595	794	6
37,	421	493	434	502	595	794	6
no.	439	493	452	502	595	794	6
3,	457	493	464	502	595	794	6
pp.	469	493	483	502	595	794	6
297–336,	487	493	526	502	595	794	6
Computer	70	496	112	505	595	794	6
Society	115	496	145	505	595	794	6
Conference	148	496	195	505	595	794	6
on	197	496	208	505	595	794	6
Computer	211	496	253	505	595	794	6
Vision	255	496	283	505	595	794	6
1999.	313	505	335	514	595	794	6
and	71	508	87	517	595	794	6
Pattern	91	508	124	517	595	794	6
Recognition	128	508	179	517	595	794	6
CVPR05,	182	508	225	517	595	794	6
vol.	228	508	244	517	595	794	6
1,	248	508	256	517	595	794	6
no.	259	508	273	517	595	794	6
3,	276	508	284	517	595	794	6
pp.	72	520	85	529	595	794	6
886–893,	89	520	126	529	595	794	6
2004.	130	520	152	529	595	794	6
[34]	293	525	308	534	595	794	6
M.	313	525	325	534	595	794	6
Hagiwara	329	525	370	534	595	794	6
and	374	525	390	534	595	794	6
H.	393	525	404	534	595	794	6
Akagi,	407	525	436	534	595	794	6
“PWM	439	525	471	534	595	794	6
control	474	525	505	534	595	794	6
and	508	525	524	534	595	794	6
experiment	313	537	363	546	595	794	6
of	368	537	376	546	595	794	6
modular	382	537	419	546	595	794	6
multilevel	425	537	468	546	595	794	6
converters,”	474	537	526	546	595	794	6
[25]	51	538	67	547	595	794	6
Q.	72	538	82	547	595	794	6
Zhu,	85	538	105	547	595	794	6
S.	108	538	117	547	595	794	6
Avidan,	120	538	153	547	595	794	6
M.	156	538	168	547	595	794	6
C.	171	538	181	547	595	794	6
Yeh,	185	538	204	547	595	794	6
and	207	538	223	547	595	794	6
K.	227	538	237	547	595	794	6
T.	241	538	250	547	595	794	6
Cheng,	254	538	284	547	595	794	6
Power	313	549	340	558	595	794	6
Electronics	344	549	393	558	595	794	6
Specialists	397	549	442	558	595	794	6
Conference,	446	549	498	558	595	794	6
2008.	502	549	526	558	595	794	6
“Fast	70	550	94	559	595	794	6
Human	97	550	129	559	595	794	6
Detection	133	550	175	559	595	794	6
Using	179	550	204	559	595	794	6
a	207	550	212	559	595	794	6
Cascade	216	550	252	559	595	794	6
of	255	550	263	559	595	794	6
His-	267	550	284	559	595	794	6
PESC	313	561	339	570	595	794	6
2008.	342	561	366	570	595	794	6
IEEE,	369	561	396	570	595	794	6
pp.	400	561	413	570	595	794	6
154–161,	417	561	454	570	595	794	6
2008.	458	561	480	570	595	794	6
tograms	72	562	107	571	595	794	6
of	111	562	119	571	595	794	6
Oriented	123	562	162	571	595	794	6
Gradients,”	166	562	217	571	595	794	6
IEEE	221	562	246	571	595	794	6
Confer-	250	562	284	571	595	794	6
ence	71	574	89	583	595	794	6
on	92	574	102	583	595	794	6
Computer	105	574	147	583	595	794	6
Vision	150	574	177	583	595	794	6
and	180	574	195	583	595	794	6
Pattern	198	574	230	583	595	794	6
Recognition,	233	574	284	583	595	794	6
[35]	293	581	308	590	595	794	6
W.	313	581	326	590	595	794	6
G.	331	581	342	590	595	794	6
Aguilar,	347	581	383	590	595	794	6
R.	388	581	398	590	595	794	6
Costa-castelló,	403	581	468	590	595	794	6
and	473	581	489	590	595	794	6
C.	494	581	504	590	595	794	6
An-	509	581	526	590	595	794	6
vol.	71	586	87	595	595	794	6
2,	90	586	98	595	595	794	6
pp.	101	586	115	595	595	794	6
1491–1498,	118	586	166	595	595	794	6
2006.	169	586	192	595	595	794	6
gulo,	313	593	334	602	595	794	6
“Control	338	593	375	602	595	794	6
autónomo	379	593	422	602	595	794	6
de	425	593	435	602	595	794	6
cuadricopteros	438	593	502	602	595	794	6
para	505	593	524	602	595	794	6
[26]	51	605	67	614	595	794	6
T.	72	605	81	614	595	794	6
Ojala,	84	605	109	614	595	794	6
M.	112	605	123	614	595	794	6
Pietikäinen,	125	605	176	614	595	794	6
and	179	605	194	614	595	794	6
T.	197	605	206	614	595	794	6
Mäenpää,	209	605	250	614	595	794	6
“A	252	605	264	614	595	794	6
gen-	267	605	284	614	595	794	6
seguimiento	313	605	365	614	595	794	6
de	371	605	381	614	595	794	6
trayectorias,”	387	605	446	614	595	794	6
IX	451	605	463	614	595	794	6
Congreso	468	605	509	614	595	794	6
de	515	605	524	614	595	794	6
eralized	72	617	104	626	595	794	6
local	107	617	127	626	595	794	6
binary	129	617	157	626	595	794	6
pattern	159	617	191	626	595	794	6
operator	193	617	230	626	595	794	6
for	232	617	244	626	595	794	6
multires-	247	617	284	626	595	794	6
Ciencia	312	617	346	626	595	794	6
y	350	617	355	626	595	794	6
Tecnologia	360	617	406	626	595	794	6
ESPE	410	617	437	626	595	794	6
2014,	441	617	465	626	595	794	6
pp.	469	617	483	626	595	794	6
144–149,	487	617	526	626	595	794	6
olution	72	629	102	638	595	794	6
gray	105	629	124	638	595	794	6
scale	128	629	148	638	595	794	6
and	152	629	168	638	595	794	6
rotation	171	629	206	638	595	794	6
invariant	209	629	248	638	595	794	6
texture	251	629	283	638	595	794	6
2014.	313	629	336	638	595	794	6
