scispace - formally typeset
Search or ask a question
Proceedings ArticleDOI

A two level fuzzy PRM for manipulation planning

01 Jan 2000-Vol. 3, pp 1716-1721
TL;DR: An algorithm which extends the probabilistic roadmap (PRM) framework to handle manipulation planning by using a two level approach, a PRM of PRMs, made possible by the introduction of a new kind of roadmap, called the fuzzy roadmap.
Abstract: This paper presents an algorithm which extends the probabilistic roadmap (PRM) framework to handle manipulation planning. This is done by using a two level approach, a PRM of PRMs. The first level builds a manipulation graph, whose nodes represent stable placements of the manipulated objects while the edges represent transfer and transit actions. The actual motion planning for the transfer and transit paths is done by PRM planners at the second level. The approach is made possible by the introduction of a new kind of roadmap, called the fuzzy roadmap. The fuzzy roadmap contains edges which are not verified by a local planner during construction. Instead, each edge is assigned a number which represents the probability that it is feasible. Later, if the edge is part of a solution path, the edge is checked for collisions. The overall effect is that our roadmaps evolve iteratively until they contain a solution. The use of fuzzy roadmaps in both levels of our manipulation planner offers many advantages. At the first level, a fuzzy roadmap represents the manipulation graph and addresses the problem of having probabilistically complete planners at the second level. At the second level, fuzzy roadmaps drastically reduce the number of collision checks. The paper contains experimental results demonstrating the feasibility and efficiency of our scheme.

Content maybe subject to copyright    Report


! #"%$'&(*),+.-(/10243'3*3!56+."7&8&
:9<;=?> @!BA?> CEDF4GHCJIKFHLDM;HCEDDF*DN
OQPRTSVUXWYSZK[]\_^a`bSVcedVU%ce[ \gfhSVZjik^almZn1RYZpo1S
qsrptvuxwytXz'{*|~}p{*'.t tXKwyz4t%ygLypt%ze%tXp%t
.tXz'{*.~yarptXz'bt%pwyz'| _XtptXz'{'
yy
t%p{*tk¡tXpwyz'| ¢p{4£Kq7¤ ¥¥
y
K¦§
t%¨.{'tX¡©.:ª«{
:ª
| |Tw6ezwy|,¬©X{%ªz'.%tyªt
®Q¯£°±!²%³,´µ±
¶·6¸º¹(»Y¼»Y½4¾K»¾¿½H¹H½4ÀTÁ.¹J¼ÂÀ¼Âà ÄXž¸ÆÁÆ·µÇÉÈ·µ¸ÊË·½ËÌXÁ¬½4ÀyÍ¹ÁÆ·T½
»¾¿Å!Î¼!Î4¸ÆÃ#¸º¹Á¸Ê1¾¿Å'¼!ÍÂÇ7¼»ÐÏÒÑ¡ÓJÔ7ÕeÖ¾¿¼Çs½HÈ(ž¿×JÁ8Å£·T¼Ày͵ë½:Çs¼Ø
ÀÙ¸ »yڵë¼ÂÁ.¸ÅÂÀL»yë¼ÂÀTÀT¸ÆÀ6ÄXÛs¶·6¸º¹a¸º¹aÍXÅÀ½EÎHÜLÚ6¹¸ÆÀ%Ä_¼LÁÈ(Åë½4Ý½4Ã
¼»!»y¾ËÅ*¼XÊË·6ÞQ¼ÑKÓaÔßÅVÖÐÑKÓJÔ¹4Ûà¶·T½á(¾¹Áë½4Ý½4ÃLÎ4Úµ¸ÆëÍ¹
¼âÇs¼ÀÙ¸ »yڵë¼ÂÁ.¸ÅÂÀxÄ!¾¿¼Ë»T·%ÞÈ·YÅ¹4½ÀyÅ'Í!½H¹v¾¿½8»y¾¿½H¹4½HÀÙÁs¹Á¬¼XÎHÃã½
»ë¼!Ê½4Ç7½4ÀTÁ.¹LÅÖBÁÆ·T½Ç7¼ÂÀT¸ »yÚµÃã¼Á8½ËÍÅXÎÒä4½Ê4Á¹BÈ·µ¸Æë½LÁÒ·Y½s½Í4Ä%½¹
¾¿½8»y¾Ë½¹4½4ÀTÁÁ.¾¿¼ÂÀY¹ÒÖH½4¾å¼ÀÍgÁ¾¿¼ÀT¹¸ÆÁK¼!Ê4Á¸ÅÀY¹4Û7¶·Y½g¼!Ê4ÁÚT¼ÃYÇsÅØ
Á¸ÅÀa»ë¼ÀÙÀT¸ÆÀ6Ä(ÖHž1ÁÆ·T½Á¾¿¼ÀT¹ÒÖH½H¾¼ÀÍgÁ¾¿¼ÀT¹¸ÆÁ6»Y¼ÂÁÆ·6¹1¸º¹ÍXÅÀ½
Î4ÜÑKÓaÔæ»yë¼ÂÀTÀy½4¾¹¦¼ÂÁJÁÒ·Y½¹4½ÊÅÀyÍvë½4Ý½HëÛç¶·Y½¼»!»¾¿Å*¼XÊË·
¸º¹LÇ7¼XÍ!½å»TÅ¹¹¸ÎHÃã½Î4ÜÁÒ·Y½7¸ÆÀÙÁ.¾ËÅ*ÍÂÚTÊHÁ¸ÅÀÅÖs¼¦À½HÈ~×¸ÆÀÍÅÖ
¾¿Å'¼!ÍÂÇ7¼»ÙÞÊ¼ÃÆÃã½ËÍJÁÒ·Y½ÖÚXèè4Ü¾¿Å'¼!ÍÂÇ7¼»Û7¶·Y½ÖÚXèè4Ü¾¿Å'¼!ÍÂÇ7¼»
ÊÅÀÙÁ¬¼Â¸ÆÀY¹½Í4Ä%½¹:È·6¸Ê·_¼Â¾¿½1ÀÅÁeÝ½4¾¸ áé½ÍEÎ4Üg¼gëÅ*Ê¼ÂÃ»yë¼ÂÀTÀ½H¾
ÍÂÚµ¾¸ÆÀ%ÄÊÅÀT¹Á.¾ÚTÊ4Á.¸ÅÂÀyÛLêÀT¹Á8½¼!ÍÞ½¼XÊË·k½Í4Ä%½s¸º¹s¼¹¹¸ÄÂÀ½Í¼
ÀÙÚµÇ7Î½4¾gÈ·6¸Ê·v¾¿½¬»¾¿½H¹H½4ÀTÁ.¹gÁÆ·T½»y¾¿ÅXÎ˼XÎH¸ÆÃ¸ÆÁ.ÜÁÒ·Y¼Á:¸ÆÁ:¸º¹(ÖH½¼Ø
¹¸Î4ë½*Ûìëe¼ÂÁ¬½4¾Þ_¸ ÖvÁÆ·T½½ËÍ*ÄX½¸º¹s»Y¼Â¾ÁÅÖk¼¹4ÅÃ#ÚµÁ¸ÅÀ»T¼ÁÒ·%Þ
ÁÒ·Y½½ËÍ*ÄX½E¸º¹gÊ·Y½Ê×!½ÍÖHÅ¾EÊÅÃÆÃ#¸º¹¸ÅÀT¹4Û¶·Y½BÅÂÝ½4¾¿¼ÂÃÆÃ½.í½ÊHÁ¡¸º¹
ÁÒ·Y¼ÂÁ(ÅÚµ¾¾¿Å'¼!ÍÂÇ7¼»Ù¹E½HÝÅÃÝ½a¸ÆÁ8½4¾¿¼Á¸ÆÝ½4Ã#ÜLÚµÀTÁ¸ÆÃTÁÒ·Y½4ÜLÊÅÀÙÁ¬¼Â¸ÆÀ
¼¹4ÅÃ#ÚµÁ¸ÅÀÛî¶y·T½Ú6¹H½¦ÅÖÖÚXèËè4Ü¾¿Å'¼!ÍÂÇ7¼»Ù¹¸ÆÀïÎÅÁÒ·kë½4Ý½4Ãð¹
ÅVÖÅÚµ¾BÇ7¼ÂÀT¸ »yÚµÃã¼Á¸ÅÀ¦»yÃã¼ÀTÀ½H¾LÅí½4¾¹BÇs¼ÀTܦ¼!ÍÝ¼ÀÙÁ¬¼*ÄX½H¹4Û
ñ
Á:ÁÒ·Y½£á¡¾¹Á:ë½4Ý½4ÃðÞ¼åÖÚXèè4Ü¾ËÅ*¼XÍÇs¼Ë»¾¿½8»y¾¿½H¹4½HÀÙÁ¹EÁÒ·Y½Çs¼Ø
ÀÙ¸ »yڵë¼ÂÁ.¸ÅÂÀÄ!¾¿¼»Y·k¼ÀÍ¼!Í!;¿½H¹Ë¹4½H¹ÁÒ·Y½»¾¿Å!Î4ë½4ÇòÅÖa·Y¼Ý'¸ÆÀ%Ä
»¾¿Å!Î¼!Î4¸ÆÃ#¸º¹Á¸Ê˼ÂÃÆÃ#ÜÊËÅÂÇ»ë½4Á¬½1»yë¼ÂÀTÀ½H¾¹B¼ÁéÁÆ·T½a¹4½ÊÅÀyÍ_ë½4Ý½4ëÛ
ñ
Á:ÁÒ·Y½B¹4½ËÊÅÂÀyÍë½4Ý½4ÃðÞÖÚXèËè4Ü¾¿Å*¼XÍÇs¼Ë»y¹7Í¾Ë¼¹Á.¸Ê¼ÂÃÆÃ#Ü¾¿½ÍÚTÊ½
ÁÒ·Y½vÀTÚµÇsÎ½H¾QÅÖvÊÅÂÃÆÃ#¸º¹¸ÅÀjÊ·Y½ÊË×¹4Ûó¶y·T½»Y¼»Y½4¾QÊÅÂÀTÁ8¼¸ÆÀY¹
½ËÌH»Y½4¾¸ÆÇ7½4ÀTÁ8¼Ãp¾¿½H¹ÚµÃ#Á.¹LÍ!½4Ç7ÅÂÀY¹Á¾¿¼Á¸ÆÀ%ÄÁÒ·Y½ÖH½Ë¼¹¸ÎH¸ÆÃ#¸ÆÁÜмÀÍ
½.ôâÊ4¸½4ÀyÊ4ÜÅÖBÅÂÚµ¾J¹HÊ·Y½4Çs½*Û
õ ö
[éWµRY÷1høåùyWµSV÷([
úaûüÙýþ.ÿ

¦ÿ

ü

¶·Y½KÄ%½4Ày½4¾¿¼Ã%Ç7¼ÂÀT¸ »ڵë¼Á¸ÅÀå»yÃã¼ÀTØ
ÀÙ¸ÆÀ%Äå»¾¿Å!Î4ë½4Ç
 !"#$&%#%#%#')(*+,+*.- */01&2
%##3#$!/%4'05*!6!-#7*.-89:;/<01*%#'=/!/:>*-4=/.:;?<@A#
'.*.&$7/*B-#+/%#'C/#D01*!6!&-#*-#89:EFG(+/*0HC'.6.I%J=/&+
:I*%K4'.3#+F!*%L*MN'.6.I%O'*&>:I*%K4'.3#+F!*%P?@A#Q*.-2
89:E/R&+/S*.%#T"&-#C*101*!6U#I%V/#ITV&+/S'.+/.="-WT
1+/*-*&?XS-89:EFD#$:FMJ&+/B%#*&S'+F 0Y349S- 5!
Q=/-#B *=*%Z.? ' ?[ +9/%#'Q'.%4=4\*-4=/.:;*.+
*#+U=/-#5*-#89:EFI?B]4*+BQ01*+/Y#;/L/:;+/*%
*(^/#B0J&%##3#$!/*.%"#$&%4%#%4'54+*.-#0IQ_ `bac?
@A#*3*%Y*U/#A01%##3#$!*%5#$&%#%4%#'C#+/*-40
4.14Z(*+/0d*(\
ïÇ7¼ÂÀT¸ »yÚµÃã¼Á¸ÅÀ»Y¼ÂÁÆ·
e#$:Ff"g=I2
h
3#%4:;\*(
Á.¾Ë¼ÀY¹¸ÆÁ
D%4
Á¾¿¼ÀT¹ÆÖH½4¾1»T¼ÁÒ·%¹
E?Uij+F&% 9(+R4&
$Qg3#-#4&k*(S/#l'.*.-4&#$&%k4I+/N#L*-89:EQ$
'.+/.=N&% N05*!6.V-WT/#)+/*-*&?im+F&%4R4!/ZS
3#-#4&n#+Q#V*-89:E)$\;(o1%pM=/&-4Q*.*%
#D4S+/*-*&:F &%#'.>'.+/.=P?q'+F#%#'B*I+F!/*.%4
+A#$&%#%4J&3#*0J&$:I&T)-WT5C0J%#43#&*%J#$&%4%#I+?
r;gs
(üTû
bt#^u
Xÿ
vM%#43#&*%j#$&%4%#%4'k4l+=$9/
IwJ:I%C*3*% S*Q(x+?C@A#$S$U%#*&B3#+4+$=%#'Q'.6.I%
/#":I*01#;yz9T{*&(A/#"01# !n4%#%#%#'l#+/*-40
_|}a&%4S4~##z/*.%4&#zwQ:;34z/<05*.U-WTD/#>#$&%#2
%#%#'1*&(G'+F#%#'1&%4V+/I'+F#%#'5*I+F!*% I?G%N% 34=2
/+$&~=3 \0J&%4#34&*%n$C(+
h
3#I%T{'.%#*+/?N]#*.+
Iy01#.[4%M/=0)-#T0J%/% &-#9T"=3 I[*-#89:EF
+Z/3#01f:I4&-#N*&(C01*!6%#'O-WTg/#I0JI6RQ$
:I*01#3F!*% &Tf%(.=-#l/*f:I*%4$I+V- *#O*.-2
89:E/B%4Z/#5*W*$E!&+/0JR%4I#l*N0J&%##3#$!/Y/#I0
4l:;3#+/+/I%Q9F!I2c*(o2/#;2&+J0J&%##3#$!/*.%4%2
%#%#' ?k#*.3U#*3#-U1+$&-#B0J&%##3#$!/*.%Z#$&%#%#+
  +01~(3#+4I+>/=%#'\*(;/3#4&%41#+/*34:;/~%4
/#\I6.I*#01%.*(^+/&$9/:C0Y3#&*%4,(*.+% 34=+/
;%#'.?"R*!6I&##$:I!/*.%n*0J&%4\34:Fg.\=34+'.:&
0Y3#$!*% ^%4)%#0J&*%)W*R-I%4;K#G(+/*0S01&2
%##3#$!/*.%"4%#%#I+\_||;ac?
ÿ%þ
t
*ÿ
WH
jüÙû
 S+%.J&'*.+401Y=*.6W%#'l6!+2
&*%4U*&(#J'.I%#+/G01%##3#$!*%{#$&%#%4%#'N#+/*-40
4b6.Z- I%k#+/**.f%f4lz/I+F!34+.?m@A4L*3*%
*(,C*'"%4{P!/*0Y- Z_|[>|.|Ea>$C*VK4+F9B#$&%nV4&
(*.+#)*-89:E[&%4V/#I%NK %4MJ
h
3#I% :;B*&(>:I*%#%#:E/
/+/%49(+%41+F&% =4&4[4&,0J/#S*.-89:;~/+/.:;
/#R4!/<?AiD%#*#+D01;4*WN4+*. *="-WTV]#I+/-4.:FN%4
A&+/+F
h
34&% p_a$\-4.={*%O6!&+/&*%4>TW%4&01$:1#+*2
'.+/0501%#'4?<@A4&3##*.+/K +/=<#$&%\0J&%4#34&*%B4&
%#n3#% :;*% 9/+/%#m:;*01*.z/n:;*%#K4'3#+F!/*.%4.:;
*(/#Q+/*-*&/)&%4{*-#89:EFI?V@A#I%O/#IT{I+F!/6.ITMI%#2
(*.+/:IN#M0J%#43#&*%e:I*%4=+F&%/1-WT0J%#'g0J&
I+3#+/-4&*%4N/*k4p4&j/*e01%#01k'.+/.34T
01*+/Z&%401*+/Z9/+$:EQI% &9Tp(3#% :E*%P[,34%./R#NK42
%4^ !LC:I*$=*%M(+5%4M01II/S4Y0J&%4#34&*%

:I*%4=+F&%/? iDW34:;%<[
3#Ff&%4vM&I+O_ }!a\4b6
#+/**.{V01;4*WL(*.+B01%##3#$!*%{#$&%#%4%#'Z4:F
$B&%L;yW%4=*%L*&(#
ñ
Í¾¸¼!ÍÂÀy½

¹

Ãã½HÈ
ñ
à Ä%Å¾¸ÆÁÒ·6Ç
_
ba
-4.=f*%k'.I%#I$:Z&'*+/#0J?miR01[D^&3#01*% k&%4
01I*.%P_ `!a4+=%./JY:;&+,&%4Q'I%4I+F& #/:;+//*.%J*&(
/#Y0J%#43#&*%M4%#%#%#'#+/*-40V?U@A#ITN'b6\*32
/*.%4B(*.+\- *n/#:Q*&(D$:I+IJ#$:II01I%F\*&(A/#
*.-89:;A%4J/#B:IU*&(^%%K4%#B;A*&(^'.+/.= I?%Q/#
:*&(U/:;+/;/"4.:;I01%.F5*&(D/#"*.-89:;[~#V-#3#2
%#'Q*(7#Y01%##3#$!*%Z'.+/#N;:;6ITN:I*01 *=
/#D#+/*-#I0 %./*\\;~*(*%=2c*&2 *%.4!/J#%#%#%#'
#+/*-401?

ÿ%û
ÿ

ü


û

Qss
ûü
t#u

%J*+F+7*5/:;+/-
*.3#+G*.+C,~K4+F=G'.6.,
h
3#$:FB*!6I+/6Wn*&(

qDv _
[|

!a ?
qDv -#3#4lk+/*..0J&*&(J%#*l%m/#p:I*%K4'.3#+F!2
/*.%p= :;

~2= :;

B*(#V+*.- *?{@P*{:;*% 9/+34:;)/#
+/*..0J&<["%340)-I+B*&(+/%4*0 :I*%K '3#+F!/*.%4R+1=I2
:;)%Y4(+/I

~2=4.:;,&%4Y&+/~34Y%#*?

*.
%#*J&+//=p(*.+Q:;*.%#%#:;*%z/*:I#%#%#+/?
:FO01VL*W:&,#$&%#%#+Y34:I:II#I[G/#N:;*+/+/ *.%4W2
%#'p'{$%4I+e% #{+*0J<? @A4"+/%4*.0
IyW4*.+/&*%N*&(G4
~2=4.:;B$D*(oI%L:I:I*014%#V-TN
#3#+/=$:54&Y4#\Iy/+/V%#*\%
"!
wQ:;3#
$#
N+/I'.*.%4
*(,/#
%
~24:I[^*l01#+/*!614Q +=(*.+0J%4:;Q%p:
#+,/#R*3*%54!/1'*W/#+*.3#'1%4++/*!g4./&'.
_
4[
&
[,|
'
!a ?l%4#6W$34&,#$&%#%4%#'
h
3#I+/)+*6O-WT
.#%#'{#l=/+=Q%4'*.:;*%#K4'3#+F!/*.%.J%#*1%
/#Z+*0Jp&%4/#I%34%#'O'.+/#+/:Fp*{K4%4f
4&Z:;*.%#%#:E%#'54U%4*W#?
@A#D##:!/*.%5*(#
(
qDv (+/01I,*+/B*\0J%#432
$!/*.%Z#%#%#%#'AR *-#)- :I&3 =\*(7#Y%/+*34:;2
/*.%*(S(34IIT+*0J4I?k]43#ITp+/*.#0145#+/*!6$Z
01#C,bT5/*1#&"#C#+/*-401A.*:;$!J
34%#'Y4+*.-4&-#=$:D*%=2c*&2 *%.A4&4%#%#I+FA*2
:&#%#%#+/[%4l*Q#+/*!6W#\;wQ:;I%C *.%2/*&2  *.%
#$&%4%#I+FI?
@A#S]#3#IT
)
qDv %./+*3 :;1%
:E*%}U*(#$>4&2
I+[b-#3##G
¡ÖÚXèè4ÜJ¾¿Å*¼XÍÇs¼Ë»
Q#*=>'P+~%#%#*/!/
-WT\U#+/*-4-#9T?^@A4#+/*-4-#9TBG%Y9/0J!/*(4/#
:F4%4:;)4&S/#5'.Y$S.:E/34&TV(-#5? ?B:;*.$=*%
(+/I.?"@A#Q4%#%#I+\#+/.9/:&TM+/#34:;{#J%W3#0Y- +\*&(
:I*$=*%g:F4:F\-WTO01%#01%4'M4"%W3#0Y- +Y*(R:F#:F
#$:Fk&+/N34=3 &TgA=#Z6+(TW%4'{'54&
+C%#*A &+A*(^/#CK4%4<=*.3#*%V4!/<?
+*
%
:E*%n`N,Q=4*! #*!H*Z3414Q]#3#IT
,
~qDv
/*)I6.I*JC0J&%##3#$!/*.%J#$&%#%#+?>XS3#+>0J%#43#&*%
#$&%4%#I+^PzF=z(#D]43#IT
-
qDv #$&%#%#+P#$:FB34(34IIT
*%=2c*&2 *%.

qDv 4%#%#I+F
!
*:&#$&%#%4I+F
.#
4?
%
:;*%
/
L,"4+=%.1*34+5Iy +01I%/,+/3#/?
02143657598;:$5<5987-=;>@?2?BACED4FHGI7.:6J9KL:MK959J3ON@PQJ97RK7RS<GTU:6N@V27L593
W
:6?BAXCED4FZY7B[\7RT]3O^_7RY`Ka]S+>T 5b:6N@723O>@KT AX:6N@YXa]NY7R^I72N@Y7RN'59T A`G'A
c
3O8T]a]Nd:6NYfeg:2[Jb:6h'ai j$kml
{\:;*.%4:;34CZ1/:;3 *.%"%
:;*%

?
n o
ø
pqp
f
srutwv
%\,+F#z/*.%4&!+/*..0J&C&%C'.7- I9I%U9,*%#*
+/I#+/I%F74D(x:;/4!A\4+=/:I3#$&+~*:I #$&%4%#I+,,.
-#R*5:;*%4%#:;~4R9,*\%#*-WTJB(-4S4!/<?7@A#
-4=%4:;J*(A%n'.J01&% C4&\4:;*.%#%#:;*%{(x&
*.+UAC%4I6+CI6%{+/
x
x- :I&3 =54J%#*B,I+/Y*W*
(x+)&4+=
O
E?N{:;*.3#n/bTM/4!Yz(/#I+/Q;y$95&%n#'
-;9,I%Z9*"%#*[#%Z4Y')(*3#%4l-WT"/#5*:I
#$&%4%#I+C$C(.=-#5{#+/*-4-#9Tg|[^&%4Lz(A%#*N#'
IyW$=/#I%"#B4+*.-4&-#z9T1$#?7@A#C#Y- #% (34IIT
+/*..0J&4^/*U%#%#*/!/,'*&(4/#A+*0J)
34:F)#+*.-4&-4z9T.[:&14
L½Í4Ä%½£»¾¿Å!Î¼!Î4¸ÆÃ#¸ÆÁÜ
!?>%4=
*(D+9/+$:E/%#'l#"#+/*-4-#9TL/*
zy
b
|{
I|
_}
,;1z5/
%TJ6!3#B%n_
~{
I|Iac#%4:;U/#B%4&01
!
=(3#IT
#
Y+/*.#01<?
@A#Q'1#+/*- &-#9Tl$\%{9/0J&J*&(,4:F4&% :;
/4!RJ !N^&+D- \(*3#%4N-WTV1*:I<#$&%#%4I+?~%
/#,:;*.%;yW,U&+/D(+/ID/*)34C\W% J*&(P*:I #$&%4%#I+[
#$:F"#*Y%#*,%4:;/&+/T1'3 &+F&%#U(-#z9TJ*(^
'.6.I%V4!/<[4-#3#D#:FZ$D-#C/*J0J&B5(x.9D9/0J!/
*(P4B#+*.-4&-4z9T*&(^/D;y=%4:;.?
.349B.5#J'.I%#+/

qDvn[4Q]#3#IT
u
qDv  C
+%#%#'M44"%4g
h
3#+TO# [7-#31%n#N]#34IIT
qDv ,,=/+=G-WT\I%/I+/%#'S4
h
3#+TC# [%4\*.%#T)z(
/#R+/*..0J&

%#$&TJ:;*.%4=%#'\*(*.%#T)/#S=/+=%4
'.*.&%4*W#
@
Q#*g%#*&V:;*%F&%fg *-4M*3*%k !P[
/#I%5893#01f/*L/#Z&+/%#%#'{#4V*O054+*!6./#
+/*..0J&<?Z]'.3#+/M|#*!D\

*! :F &+\*3/%#%#'M/#
'.*+/z/#0N?\%l/#J]#3#IT
u
qDv ,J*V%#*&B+/3#%l/#1*2
:&P4%#%#I+R*Q6.I+/z(TV#'S##Z3#+/%#'J4)&+/%#%#'
# ?>%4=V*&(76.I+/z(TW%#'5#\'I[,\/'.%"/#I0
%p%#z/'.N#+*.-4&-4z9T.?p@A#%$!/I+1%/#
h
34I+/T
# [P(R&%O'J$\4+=\*(DN *-4*3/*.%{ !P[
,B&##TJ/#B*:&#$&%#%4I+A*J34#'+FU/#+/*- &-#9T?
ÿ

ÿ%û

s

t
|
'ÿ
@A#
h
3#I+/Tg#4.=N$Q#Q%./*
/#+9/I4

^%Y3##&[WC=&+F:F<[&&% Y%Y3#4'+F=IP?
%{#13##&Q=IO,J#L%#*C+/I4+=%./%4'/#
=/+=Y%4O'*&~:I*%K '3#+F!/*.%

xz(/#ITn+J%#*5+/.T
/#I+/
@
,&% V:I*%#%4:EA/#C*54B+*0J&P?
%M/#Y+/:FM=M,\K4%4M#501*.=S#+/*- &-#\4&
(+/*0 9F&+\*l'*.74+*.3#'O/#+/*..0J&<?Z@A#Q4+*.-2
-#9Tg*&(BO4!/

$QIK4%#k54M#+/*W#34:EQ*(S/#
#+/*- &-#U*(~45'
f
*
{._{'{$@
l&*%#'"#J !
4
~
*

E?
{NK4%4#$14&-T=I=%#'O/#N'.V,I'F)/*
9;9@E
9.
&% N#%Z-WTV3 =%#'S 89=+F
R'.*&2
+/z/#0H*\K %414R4&QJ/#R01%#0)3#0 3#0 *&(<#'
,I'/?k (C%#*n4!/f;y=Q,V+/;/3#+/%(x3#+/[,I$=N,
#+/*:;"z/"/#B3##'.+/.C9/I<?
@A#O3##'.+/.L= 4%4"4O.:E/34&U6+K :I&*%
*(/#Q4&<?Z]+F=\Q%4I+)&~'B%/*MN#+*+/9T

Return path
NO
YES
Query phase
Update step
Upgrade step
Remove
edge
Path probability?
Search step
Path found?
Construction step
Expansion step
Learning phase



0
]'3#+/)|
G]*!e:F4&+V&%Q*!6.I+/6W*(<(3#IT

~qRvn?
h
3#3#Cz/"/#I+#'C#+/*-4-#9TQ#+.IT?G@A4I%
,>+/I!P#~(**!%#'**.B3#%.z/#I+&'<+/:F
#+/*- &-#9TQ|*.+>3#%#/#*:&##$&%4%#I+~#/:;*!6.I+FB:I*2
*%M&*%4'1#\ !
4<
>y/+/.:E/#)'BZ/#\*!A2
9\#+/*- &-#9TZ(+/*0 #
h
3#3#[^0J&1/#J*W:&>4%2
%#+7 +=(*.+0H#*% &:;*.*%1:F#:FG&*%#'S4, !P[
%4n/='%L4#'QZ%# #+*.-4&-4z9TLII%4#%#'L*%
/#D%#e:;**.%J:F#:FY+=*.3/*.%<? (/#R*:I##$&%#%#+
$/:;*!6+/7Y:;*.*%<[.R#I;R#U:;*.++/ *.%4%#'B#'
(+/*0 #)+*0J&Z%4N+I3#+/%N*Q#)+/:FZ=<[ 
,)+/I%4I+D#Y'.)%*Q/#)4+*+/z9T
h
3#I34Y&% M:;*.%2
/%W3##"3##'+F#Q*W*.<?M (4#+/*-4-#9TL*(R&>/#
'&*%#'1/#B4!/ZD3##'+F#*Z|C4I%N4
h
34I+/T
$=*.6.V&%4",B+I3#+/%/#B4&<?
k#%n%#I{*M3#4'+FQZ !n,:&7#(*2
*!%4'5*:I#%#%#+


! "$#&%')(+*
[
 -,!./'10*2,
E[

3

465
-,!./'10*2,

7 "$#&%8'9(+*
:;
<:=5
}
?>
*+@*
>BA
*
C
5


þ.ÿ
CED
:-;
F:
;E5
"$#&%8'9(+*HG
'
)I9"2,!*KJ-"
G
'
"2,!*KJ-"
C

L
;M
NPORQ
')*$*
ûÿ

û
!
t
9
*ÿ
C
5
C
G
e|
ûÿ

û
!k
ÿ
@A#Y*:&#%#%#+UI+(*+/0JR:I*$=*%L:F#:FS*%Z/#
=+F&'%#C- I9I%Q/#B=*.3#+F:;C&%4Q/&+/'I,:I*%K4'.3#+F!2
/*.%l*&(~#14&l.? c?C45%#+S%I+/*$!*%l- I9I%
"$#1%8')(+*
&% 
S ,!.'10*2,
E?>@A#SK4+F9,01S,U%#IJ*53##'.+/.
V4&OJ:&7#J*:I7#%#%#+Bz/
,9
3
<
|[P%4
/#1*:I7#$&%#%#+B +=(*.+0JBV=%#'.J:;*.$=*%O:F#:FM&
/#R01$#R*&(/#R4&<[.4I+/I-WT5#6W$%4'\4R4&J%./*
9,*n3#%4:F#:F:;*%4?jDIy05Ll%#I*g3##2
'.+/.\/#J4!/L,5%4:I+
9
3

>-WTM*.%#1&%4{:IG/#
*:I&~#%#%#+Y&'&%<?N@A#Q+/3#zYC/4!5J &6J/#
$=/&% :;C- I9I%Z34:I:I/=6C:;*.*%N:F#:FA&*%#'5/#
%4U:F"01BB3#4'+FS#B'.?
{~4b6.>:F#*.I%C/*D0J&>#,=01#(TW%4'S/3#01*%
/4!,#C%W3#0Y- +*(P01AY'6I%
%
~2 4:IR4&"4/=
/#J- *.3#%4#+TL*&(,#
%
~2c*.-49F:I1*%4T{II%44\*%{/#
I%#'*&(U#Z !P?kLM0J"#$Q/3#01*%fI6.I%
/#*3#'.CzP%#*&</+34G(*+P01*9^=+/34:;3#+/UI%W6W+*.%#01I%/[
-:&34lM&*!D34"*f:F#*W*.l054{#+/*-4-#9T
01*Ic[+&%#'J4)I%#'Z*&(7/#) !M*J/#)%W3#0Y- +
*(*.-49F:I7-*3#% #&+/TRI%#I+F!/*.%4I[%405TS4

G*$*%
$=+/-#3#*%
7T
y
U
z99
WV
}x
AYX>?Z
X>\[^]
`_
[R#I+/
B$
/#"I%4'&p*&(D/#" !P[
aU
z
B$1l=*:F4=$:Q6!+$&-#
+/I#+/I%/%#'\#U%W3#0)-I+,*&(/01/#S !V:;+/*./=7/#
~2 *-4=/.:;7- *.3#%4#+T.[!&%4
Lb

#$P#05&%\6!3#*&(
U
z99
;?
( R4!/\*&(#I%#'
Q
44^-II%Y=34::;=(3#TB:F#:FB&
I6.I
~9
3

-WT)/#R*:I&4#%#%#+~*3#%#J&-*!6.[&4I%Q
4.- I%16W$Y%./*\}
?>
*+@*
>
3#-4:;*%4[!.:F)*( I%#'
>
)cd!e&dfc
?GT)34=%#'S#

G*./*%B#+/*-4-#9TB01*W#I#&% \/#
W%#*!'.B !
gU
0)3 9C- 5I6.I%
`
x- :I34=Y45:I*%K4'2
3#+F!/*.%g%:Fp%4*&(D/#N4!/$1:;**.%g(+/I

\,
:&%l34:I\4)(*.*!%4';y#+//*.%N(*+D/#5#+*.-4&-42
9TJ !R54!/N,(-#Q_|`ba
b
;
R{$
3
@
ih
_
j
'
:k
fb
>
)cd!e1dKc
a
A
cdfe1dfc
>
)cd!e1dKcml
<:
|
>
cd!e1dKcmn
<:
{
9|

#+

F:
$J/#O=0J&="$9F&%4:I[De4:Fk,
A&%1*nI+(*+/0 :;*.$=*%f:F#:FJ*.%#'O4!/4I?D*&/
/4!BC$C*./=-#Y/*V:F4*.

<:
6I+/TZ=0Jc[P%4:I1zB$
3#%4TV !U#1I% =5+=*.3#*%ZG-Y+/.:F#N(*.+
%TJ !NIy:II/#\*3*%V4!/<?
b
f:I%p-N=01&g;yI+/01%.F&Tn3 =%#'L#V(x:;
/4!
mo
Q_
U
z99
a
q
pb

9?>,T":F &%#'.%#'Y4
Lb
Z4+/05I+,D$
*./=-#,*\:;*.%+/*W4#%#%#+/G#+/;(+%4:;A/*!,+/\(x=
#$&%4%#%4'"6?\=#*.+=C=*.3#*%l ! I[-#3#U/#$R(!/3#+5$
-IT.*%4#\/:;*. C*(^/#A &I+?
q
éÿ
t
bx
r
1s

^t
|
'ÿ
ig:;*.%49/+3 :E*%B=B%4I+/
U
+/%2
*.01TU:F#*.I%C:;*.*%U(+7+*.- *P:;*.%K4'34+/&*%4<%#*
%*A47(3#ITS+/*..0J&<?P <#%\:;*.%#%#:;/:FC*&(W/#
aU
%#n%#*^/*R/
Hs
:;*.=^%#'.W- *.+/Pz/)#'?
:F
*(,/#Q'B+#+/I%/BZ !n-;9,I%O9*Z%4*W#?
t
R%#A%5#'.I%#+/

qDv *C%#*>+34%5U*W:&#4%2
%#+U*N6+(TN#J'.[% 9/L,1.'%l\&%{%4z/
#+/*- &-#9TJ'6%"-WTQ
h
34!/*.%

=|
@
,

3
@
e#?
iD%1IyW &%4*.%5=

x01+7%1%4&3#+/,*Q_
ba
m
>:F#*W*.
u
%#*\&% O +=(*.+0J\N+F&%4#*0 A&L=/+=/%4'"(+/*0

Goal
Start
]'3#+/S}
Gi01%##3#$!*%J'.+/#<?
*$J%#~+/I#+/I%
/+/%4=P4&4&% B*&\%#^+/I#+/I%+F&% 9(+P4&4?
:FZ*&(G/#)%#*I?

:FN*&(7/#
u
%#*S&+/\:F#*.I%
+F&%4#*01Tz/N5#+/*- &-#9T
!3


h
34&/*
~
^3

C

:
!3

}

;
;

C
9
:
{
}
#I+/
C

:
^3
VN/#g%W3#0Y- +M*&(1'.M=/+=/%4'e!
%#*
m3
\ !~,I+/AK4+F9.#5*B/#D+/*.#015%41$!+
Il3#+/%#'Q/#
h
34I+/TV44?
;
;

C

:
$S#Y*2
F&#%W3#0Y- +>*&(I;/5#'?%1:
;
;

C

:
>C,
W"#BIy4&%4*%N9/I<?
o
ø
pqp
f
rutwv
÷¡R
wv
óZK[åS

åøådZWYS÷([
XS3#+"&##+/*..:Fp/*g0J&%##3#$!/*.%k#$&%#%#%#'g${9,*
I6.I&01;/#*?P%C/#7K4+F=<6I74b6.7,(3#ITR+/*.#01
#$:F4bT~#C+/*R*&()0J&%4#34&*%'.+/#l_ `!a ?7@A#
#34+*.7*&(W4>0J&%##3#$!/*.%B'+F&#C$*R#:;*.01 *=G/#
0J&%4#34&*%k#$&%#%4%#'p#+*.-#I0 %*p0)3##l*%.2
/*&2  *.%A4!/"#$&%4%#%4'54+*.-#0JI[W#$:FV+C*6-WT
(3#ITU *.%2/*&2  *.%
4
~qRvH4%#%#I+F<&<#=:;*% SI6.Ic?
:FZ%#*Y%Z/#501%##3#$!*%l'+F&#Z+/I#+/I%FR
$&%4#01+N:;*.%4=%#'J*&(~J+/#Y:I*%/%4%#'"1+F&%4=(*+2
0J!/*.%0J&+/zy

#

[B'+F %W3#0Y- +
PV
0)')."fJ
[B&%4
+/*-*&1:;*%#K4'3#+F!/*.%n6:E*.+
'9#

?

#

 :;K45l=/&2
-#N#$:;01I%5*(R/#N01%##3#$!g*.-89:;[~4I+//#
*.-89:;):I&%{-JI(o)3#%#'.+/.=L-WTM4J+*.- *?
V
0/'9.-"fJ
$
/#B%W3#0)-I+*(G5'+F+/I$!/*.%*(^/#B*-89:E?Gim'.+/.=
+/I$!/*.%e=:IzK4nI'&S *=*% %4f*+/I%/&*%f*&(
/#5*W*Gl+/:EU*"454.:;I01%.C*&(~#1*-#89:E[
&Q#:F/#N*W*RQ-#N*n'+Fp*.-89:;?
p '9#

1$Q
:I*%K4'.3#+F!*%56:E/*+G(*+>#D+/*-*&[!(*+~#:F5/#D+/*-*&
/**.$)-#Q*L'+F{#"*-89:EY4.:;p!

#

3 =%#'
/#\'+F":;K4N-T"'+F"+/I$!/*.%
VY0/')."fJ
#?@A4B01&2
%##3#$!/*.%p'.+/#p')+/I#+/I%F\/+/%4=J&%4p+F&%4=2
(+S:;*%4?AP&% 0J&+/WD#*.B'+F"+/I$!/*.%4D+C/#
/&01Z&+/V:I*%#%4:E/-WTpL+F&%4=(I+1#'?k^&%40J+
#*=U*-#89:EA*.z/*.%4A&+/R/#B05C&+/S:I*%#%4:E/"-WT
Y+F&% ='.?
]'3#+/1}"#*!DSN0J&0J%#43#&*%L'+F&#<?YD*&/
/4!19,*L*(R4V$&% 0J&+/W1+V:;*%4%#:;-WTg-*&f
/+/%49(+\&%4OV/+/%4=B'.?@A4B$B-:&341#19,*
$&%4#01+>4+-*&/J#U01D*-89:E+F&%4=(*+/0J!*%
%4Q/#\/&01C'+F%W3#0)-I+?
(#,74B:I*01#;~4!/\#$&%#%#+/P!</#~:;*.%4CI6
,%O_ `!a

E[,S,*3#$"- B&-4S*16+(T14UIy=%4:;U*&(
:FN*&(G4B+F&%4S&%4V/+/%4=(I+D4&4R%N/#\0J&%4#3#2
$!/*.%M'+F&#M#3#+%#'V:;+/!/*.%<?UL1:;*.3#Z#1=*.6.\/#
0J&%4#34&*%M#$&%4%#%4'"#+/*-40 -WTM.#%#'V.#*%4
$&%4#01+U+/I#+/I%%#'#Q=/+=\&% l'.*.&7:I*%K4'.3#+F!2
/*.%4*&(4~=T=0V[!%4C=&+F:FR/#>0J&%##3#$!/*.%C'+F&#
(*.+YZ4&p:I*%#%4:E/%4'Z#V9F&+Y&% n'*&>$&%4#01+?
]%4&TO,:;*.3#$g:;*.%4:I&I% !Q#V% 6W$34~/+/%49(+
%4Y+F&%4z'.>*.%#'C#R4!/J%*\#R*3/*.%J01&2
%##3#$!/*.%" !P?
i wJ:I3#C4+*.-#0 S !B534Y*.%#TV4+*.-4&-#=2
/:&Tn:;*.01#I"#%#%#+/Y!)/#":I*%4g6Ic?
~+*.-2
-#$9/:&Tk:;*.01#IO#$&%#%4I+F"+{3#%4-#L*f:;$
#I#+\N'6%
h
3#+TM$B3#% =*.6!&-4
C#Tl>%4I6+
/I+/05%4&~(
h
3#+U(*.+%\3#%4*6!&-#~4&<?G@^*R34>/#
- *!6.\&'*+/#0 ,\%#IZ/*":;$)*.%l=*.01)0J!y01
+/:F1/01[#34:FJ/4!A(P4U:I*%4JI6.I#$&%#%#+3 =
01*+/Q/4&%p#$1&01*34%.5*&(D/01"#%p,V#:;$"4&
/#
h
3#+T3#%4*6!&-#?,3NpI2cI:E":I*3#$k-
/4!~wQ:;3#~'G#:FJ+A%#:;//&+/T\%5/#R*3*%
4&N&+/C+*.%#'&(3#TI;/"(+/*0 /#B+*0J<?
{)4b6.):F#*.I%ZQ#zI+/I%U##+/*..:FN%4N,\34)
(3#ITL+*0J&{/*Z+#+=%B#Q0J%#43#&*%O'+F&4<?
{>;P/#~'~#+*.-4&-4z9TS*&(#'.6.I%C'.

,II%4\*%
#*!n0Y34:FB/01#A:;*.++/ *.%4%#'R:I*%4BI6.I#$&%#%#+
4.=I%R+/:F#%#')(*+A4B4!/V+/I4+=%./-T

9;
m|
L
;
C

;
;
;

;
C

{
x`

#I+/
;
C

;
RU/#101Q=I%B-WTZ#Q:I*%4LI6
qDv +/:F#%#'n(*.+V#{4!/+/I4+=%./k-WT
[S%4
;
;

;
C

JB#1/*&/G01Q=I%)-WTl>:I*%4{I6
#$&%4%#I+FI?
@A#V0J&%##3#$!/*.%g#$&%#%#+Y$5&'.%6W$g%*{
+%#%#'Y#4C&%4"
h
3#I+/TJ#4.=.[W#:FV&+/U#'.%#
-WTZ&%4*.'T"*"#1:;*.++/*%4%#'Q44U*&(
:;*%L}#[
Iy:II^4&P/#0J%#43#&*%B'.+/#B^-#3#$)3#+/%#'D/#
%"%#$&<+/3#%V*&(#B+%#%#'5#4.=.?
ÿþ.ÿ
t
bx
r
{s

t
|
'ÿ
%"/#\:;34++/I%6+/*.%*(^/#
#$&%4%#I+U453 =+B*+U#J:II%\=*(o9,+54.R/*Z=3#4#T
-*&/

#

[
VY0)')."fJ
C&%4
-')#

G(*+G.:FY*(44%40J&+/^%
/#Y0J%#43#&*%{'+F&#P?B U$C *-#Y*V+/34:I)/#$
+/
h
3#+01I%*B893 9

#

B-WTN&3/*0J!/:&TV:I:I3#&%#'
'9#

%4
V
0/'9."KJ
3 =%#'D%W6I+F7%#01&$:IP*+^D%#01&$:
+/*..0J&_|Ea ?
#*W*.%4'"/#Q=/&-41*-#89:EB*.z/*.%4C$
"4&+FM4+*.-#0%{/I(9?Y{J:;3#+/+%TV+/ITl*ITN*%
/#S3 =+/*5.#3wQ:;I%9F&-#U *=*%4~*50J&.R/#
#+/*-40=*.6!-#.?
@A#"%40J&+/B*&(*34+)+/*.#01n&+/Q#O*%#-T
*.%#[)%4j:Fj$&%40J+{:;*.%#%#:;*e/#p*#+

$&%4#01+J%e/#l+/*..0J&<?j (B9,*g$&%4#01+4b6
/#Q015*.-89:;\:;*%#K4'3#+F!/*.%M/#IT{&+/5:I*%#%#:E/{-WT
N/+/%4z\'.?V (A/#IT{4b61/#"/&01J'.+/.={+&*%
/#ITM+5:;*%4%#:;l-WTN+F&% 9(+S'.?
M
~[%#*V#'
$S.#?DiRS%M#5]#3#IT

~qDv ,)*"%#*&S6I+/(T/#
(-#z9TJ*(<#C'4I%Q/#IT"&+/S:;+/&[W%4=
%#'G&+/,.'.%#5&%5#'A#+/*-4-#9TB*&(
~ _
=%4:;J|
Y01%4A !A#B !N4.A- I%V(*34%4
~
E?
%e:V*(\O+'.Z%340)-I+Q*(C$&% 0J&+/W1"01'
-\%#:;//&+/T1*J*%4TV:;*%4%#:;D:FN$&%4#01+Q*J/
s
:I*=9%#'.-*+F[W:F4*.I%V-T"=*.05B$=/&% :;C01;+/$:&?
ÿ
ÿ%û

s

^t
|
'ÿ
{l=/+=J/#
h
3#I+/Tp44Z-WT
3##&%#'N#J0J%#43#&*%O'+F&4L{4Q9F&+)&%4
'.*.&G:;*.%K4'.3#+/&*%Z*(>#1TW=0N?U (#5*-#89:EC$R3#%#2
'.+/.=O%pz/#I+Y#N=/&+Y*.+Y'.*.:;*.%K4'.3#+/&*%O/#
:I*+/+=*%4#%#'Y%40J&+/QP4b6C&%V3#%4IK4%#"'.+/.=
%W3#0Y- +?%N#$D:I.=C/#\%#*BP*%4T"- ):I*%#%4:E/
/*L*#+Y%#*5z/p/#V/&01"*-89:E5+F&%4=(*+/0J!*%
-WTJ/+/%4=D4&4?
@A#R*:I 4%#%#I+F>3 =J3#+/%#')#S3##'.+/.R=IQ*
K4% M(-#\/+/%4zC*+S+F&%4=(I+U4&4S(*+U/#Y%46W2
34<#'*.%#'1101%##3#$!*%N4&<[ &+/C#zI+/I%D%2
=/%4:;7*&(#D(3#IT)*%.2/*&2  *.%

qDv #%#%#+#XS%#
%4=/&% :;Y(*+U:FL *-#Y'+FM*&(~#1*-#89:E?\@A#
#$&%4%#I+FC&+/)34M*N#$&%M/#5+F&%4=(I+U4&4?)@^*"3##2
'.+/.)1/+/%49(+S')z/LQ &+$:;3#$&+S'+F<[4,)34
/#\]#3#IT
~qDv /=*:I&z/V#$'+F<?
{l:;*.3#$ b6N% :;34k%k% 9F&%4:IZ*&(U#Z(34IIT
*%=2c*&2 *%.
g
~qRv #$&%#%4I+(*+G.:FY=/-#,#$:II01I%
*(<#C*.-89:;[&%44I%V34=S4S#$&%4%#I+F*5#%Q/#
/+/%4=\4&4?
% :;J/#Q%W3#0)-I+Y*&(D=/&-4Q:I*%K4'.3#+F!2
/*.%$534=3 &Tn$&+/'

$&+/'I+Y4I%#N%W3#0)-I+1*(S(o2
(+%.\'.+/.= 
$
;[P&%4n%4:;J\)3434&Tl0)34:Fn+C*
#$&%nV/+/%4=\4&{4%{*Z#$&%nV+F&% 9(+B4!/<[^,
4b6.:F#*.I%1B+%.,=*.3/*.%<?L34RC%#'.A(34IIT
*%=2c*&2 *%.
(
qDv *Q*6B&/+/%4=D4&4?>{)*
/#D-T"/=*:;$!/%4'51=*&(G#+/*-4-#[*%#U(*+:F
=/-#)*-89:ER*.z/*.%<[z/Z#)'.[4%49/.Z*&(8934=
*.%#\#+/*-4-#9T?A{Y:F#*W*.C4)+/'..SI%+/T"%N4)=
 %4%#'Z*%L4Q9F&-#1*-89:E\#$:II01I%\/=*:I&
V4C+F&%4z4&N,C+B:I*%4#I+/%#' ?
'#$:Fl&+/\/='%4V#+/*-4-#9TV1(*.+SJ'.6.I%
*.-89:;M#.:;05%Z+V8934=l#$#%m#I%m,n01n
h
3#+TO(*.+14&14.:;I01%.?@A4"'J&+/"#*!,I6.I+
.I%Q4U+/*..0J&"3#%#T&+/S.'.%##+/*-4-#2
9TJB(*+A& *-4S*-89:E4.:;I01%.FI?G@A#U#+/*-4-#2
9TJ*&(</#C'.%"4S0J&%4#34&*%'.+/#Q$,'6I%"-WT
h
3 !*%
z
`
;?A]4+*.0/4!U
h
34&*%NzC:I%Z- )I%N4&
%B'.>%#I6.I+P-.'%#S4+*.-4&-#z9TC#?@A#I+/;(*.+
,%#6+5IV'.)(+*.0 #V0J&%##3#$!/*.%p'.+/#<?
34# *=A/4!7/#34I+7*.+G#D:II%~*&(o9A&+/4.>4
3wQ:;I%L%4*W#l#3#+%#'k/#p+%4%#'e= 34:Fm4&
/#I+/Q;y$95M4&
4

$
B(+*.0 =/+=\/*M'*&,&%4n&)=
*.%#S*(<4S4&4,$:;34T5(-#.?>@A#I%"*3#+'.*&2
+/z/#0 *.%4:;C%
h
3#+T#4[DP=/bT1/#I+/B3#%/#
0J&%4#34&*%V#+*.-#I0 $D*6<?
i

gceRTS

ce[éWµZKd
t
ïceU6ødWµU
@A#~'.*+/z/#0eA01#01I%C%
)
GgG
n*%\
<
..v

R
G%./340=A+/3#%#%#%#'Qk%4*!DSR@C?XS3#+D#+/*'.+/0 34
|

`l:I*%44Y*.%g/#Zb6I+F&'.
u
9|l3#-4
h
3#%.J+/3#%4
6
B/*
*6O/#p0J&%4#34&*%j#+/*-#I0 #+/*-#I0 #*!%m*%
]'3#+/Q`#?V@A#)#+/*-#I0 %W6*.6.B%O<2 4 L*.-89:;
%45D%#$&T54.C9F&-##$:II01I%/~*&(4D*-89:E?
@A#U0J&%##3#$!/*.%"'.+/#":I*%/%4)*/*&(|

Y$&% W2
0J&+/I?
Rb
ZA9/0J&V/*
|
}}#?@A#)'*.D*Q-4+%#'
/#*.-89:;)(+/*0 #V*.z/*.%p=4*!%g!5%44=4*&

;[
#+5/#J*-89:EB$C- #% {#J+/'.U*.-49F:I.[*N/#
K4% & *=*%N#*!%"&%44=4*&
f

m
E[##+S4U*.-89:;
$B-;9,II%{/#J9,*Z*-4=/.:;C*%O/#J;(o?Z,:&341*&(
/#S$&+/'S=IU&%4Q#C9/+/%#'S=4 S*&(</#U*-89:E[/#
+/*-*&14Y/*{431#N*-#89:EQ*!%&%4p+'+Fg5/*
:I*01#;U/#C/.=?
SVU6ùøåU%U6SV÷([
{S4b6.R34:I:I/9(3#T5='%4Q%4J05405%&%
'.*+/z/#0:I4&-4B*(G &%4%#'J0J&%##3#$!/*.%V4%#%#%#'
(*.+PDTW=0:I*%4$9/%#'*(#D=%#'.>+/#3#%4#%.<+/*-*&&+/0
0J&%4#34&%#'Y5=%#'.S05*!6!-#S*-89:EAz/")K4%#C=I
*(P9F&-#S4.:;I01%.FI?7@A#$,,.#*%#S-TJIyWI% %4')/#
qDv (+F&01*.+Nn&%{'.Y4+*.-4&-#z9Tl%#%#*/!/
+/*..0J&<[:IVY(3#IT+/*..0J&<?
ûÿ
)ns
ûü
ÿ
W
ÿ
W

S3#+/%4'#A6I*405%P*&(
/#):;34++/I%S##+/*.:FP[\ b6\$:I34/=V#Y *-42
9T1*&(P01*!6W%4'\4S/+/%4=(I+,4!/4>(+/*0 /#C=:;*% QI6
qDvM~*Y#SK4+F9A6 0J&%4#34&*%'.+/#<?G@A#$
+/I01*!6)#1/011*.=C4I%
h
34I+/T%#'V41:I*%4LI6
qDv (*+5/+/%49(+Y ! )/4!J*{%#*&1;y$9?p]#34+=/#I+2
01*+/[G%:Q",%)*L'%#I+F!/J#V9F&-#*.-89:;
:I*%K4'.3#+F!*% ~&3#*0J&$:I&T[zA- U*./=-#R/*YI2
3 :;B%(*+/0J!*%V(+/*0#\0J%#43#&*%N'+F&4V- *.3
#+1%l/#J,*+/W4.:;)\,*3#$l-Q#6b%/&'.I*.34D/*
'.I%#+/&Y#=Q:I*%K4'.3#+F!*% I?J@A#$C#N$C3#%4+\%#2
6.='.!/*.%<?
®Q´

±X°
 
% #&
"!
p4
$#%!'&
4&
)(
µ¿8)º*
$*
g
%!
%
,+
#
.-/
X #&
10
2&
*"
43
£ #
5+
#&8"
46
 #&
17
80

90$
Xa
:<;
!&8 #
=&>+ -

?*@(AB0
 #"X$
DC&>E
µ*
&F0
*
G;&>0
-( Ë
IH
£"X
26
'Ë
G7
&8
80;KJ L
¦*
$#
B'"
D0$
% #&
(M&?(
µ¿
NEK;OCP;
XX
2&IQ"&=6

&?#
Â
&
*&
E
µ"s&
GA(R7
(
µ*
G0
 #"
%(&
4
G0"EK;S3
10
RTU- V
é-
WW
,-
XV!'&
HV+.-é+
Y7Z[
*3'2'2
?\O&
*"6
3
10
RT.-
+¬0
BW
s0
RV^]=[
'2
?\F7
2
<]]
2'2
F3)J? 
%
_&?AR0$
%4&
P!

A+

`+
#
,#

a0

0$M&
*"
B#
0$
%K&
10
K4)
b0$
X£
:K;
!&8 
5&?+P-

*@(BAB0
 #"%$
^c
é
?A(.&>0
-é #
H
£"X
26
'Ë
G7
&8
80;
5
ed
é"X
,#fc
£*V!'"E
g0
Ë&8Ë"Ù5
RC&
H&p/
N6
'Ë$
e&?&
4V
@&
*"%
@h
L*
G0
Ë"
C
Ù #"6å
g0
Ë&8"
I&>0_H
£"X
26
'Ë&8
80;
*)Ù0Â
AR0$
%Ë"
^i
:"
B*j&
4
$#
:)º4
0$
X ð

*@*
gË"
<0
&
&
*"6B&
GA(B(
µ*
G0=J

Citations
More filters
Book ChapterDOI
01 Mar 2010
TL;DR: Although some elegant complete algorithms exist, their prohibitive computational cost has motivated the search for efficient algorithms that can run in significantly less time despite the fact that they are not complete.
Abstract: Planning the motion of a rigid or articulated object among static obstacles is known as the basic motion planning problem. In its simplest definition, it is a purely geometric problem that only considers avoiding collision with static obstacles, given that the robot is the only object that is able to move in the environment. This problem has been extensively studied in the last two decades, and we do know that the problem itself can be very challenging. For instance, it has been demonstrated that planning the motion of a set of polyhedral objects forming an articulated structure in an environment with static polyhedral obstacles is PSPACE-hard (Reif, 1979). That means that although some elegant complete algorithms exist –the most efficient complete and general algorithm for basic motion planning has a O(2n) complexity (Canny, 1988)-their prohibitive computational cost has motivated the search for efficient algorithms that can run in significantly less time despite the fact that they are not complete.

6 citations

01 Jan 2018
TL;DR: The approach leads to the synthesis of a robust real-time planning module that allows a UAV to navigate seamlessly across environments and speed-regimes and establishes novel connections between the disparate fields of motion planning and active learning, imitation learning and online paging which opens doors to several new research problems.
Abstract: Mobile robots are increasingly being deployed in the real world in response to a heightened demand for applications such as transportation, delivery and inspection. The motion planning systems for these robots are expected to have consistent performance across the wide range of scenarios that they encounter. While state-of-the-art planners, with provable worst-case guarantees, can be employed to solve these planning problems, their finite time performance varies across scenarios. This thesis proposes that the planning module for a robot must adapt its search strategy to the distribution of planning problems encountered to achieve real-time performance. We address three principal challenges of this problem. Firstly, we show that even when the planning problem distribution is fixed, designing a nonadaptive planner can be challenging as the performance of planning strategies fluctuates with small changes in the environment. We characterize the existence of complementary strategies and propose to hedge our bets by executing a diverse ensemble of planners. Secondly, when the distribution is varying, we require a meta-planner that can automatically select such an ensemble from a library of black-box planners. We show that greedily training a list of predictors to focus on failure cases leads to an effective meta-planner. For situations where we have no training data, we show that we can learn an ensemble on-the-fly by adopting algorithms from online paging theory. Thirdly, in the interest of efficiency, we require a white-box planner that directly adapts its search strategy during a planning cycle. We propose an efficient procedure for training adaptive search heuristics in a data-driven imitation learning framework. We also draw a novel connection to Bayesian active learning, and propose algorithms to adaptively evaluate edges of a graph. Our approach leads to the synthesis of a robust real-time planning module that allows a UAV to navigate seamlessly across environments and speed-regimes. We evaluate our framework on a spectrum of planning problems and show closed-loop results on 3 UAV platforms a full-scale autonomous helicopter, a large scale hexarotor and a small quadrotor. While the thesis was motivated by mobile robots, we have shown that the individual algorithms are broadly applicable to other problem domains such as informative path planning and manipulation planning. We also establish novel connections between the disparate fields of motion planning and active learning, imitation learning and online paging which opens doors to several new research problems.

6 citations

01 Jan 2015
TL;DR: A novel motion planning system that utilize several locomotion generators in order to realize basic motions of the robots — motion primitives is proposed and the proposed planner then constructs the plan using these primitives.
Abstract: The thesis deals with the motion planning problem. In this problem, the task is to find a path or trajectory between two places in a known environment. Motion planning is mostly studied in robotics, but its applications are far beyond robotics in areas like computational biology or surgery. A wide range of motion planning problems can be solved using the concept of configuration space. Due to high number of dimensions of the configuration space, that is equal to the number of degrees of freedom of the robot, it is not possible to discretize the space and search it using standard state-space searching methods. Sampling-based motion planner like Probabilistic Roadmaps of Rapidly Exploring Random Tree solves the planning problems by randomized sampling of the configuration space. A well know bottleneck of the methods is the narrow passage problem. In order to speed up motion planner and to increase reliability of the planners, we propose to utilize the knowledge of the workspace to help sample the configuration space. The knowledge is represented using a path, the guides the sampling in the configuration space from the start configuration to the goal configuration. The guided sampling is studied in three challenging scenarios. The basic principle of the guided sampling is introduced on the example of motion planning for mobile robots, which requires to sample the three-dimensional configuration space. The low dimensionality of the configuration space allows us to compute the guiding path as a geometric path in the workspace using standard path planning methods. A different approach to compute the guiding path is proposed solve the path planning problem for 3D objects, that requires to search the six-dimensional configuration space. The proposed method first solves a relaxed version of the problem by scaling down the geometry of the robot. The found solution is then iterative improved until the solution of the original problem is found. Finally, a novel motion planner is proposed for motion planning for modular robots. Modular robots are formed by connecting basic robotic modules. These robots can be reconfigured to various shapes and they represent systems with more than 6 degrees of freedom. Motion planning for modular robots is challenging also due to necessity to control many actuators in order to achieve a motion of the whole robot. We propose a novel motion planning system that utilize several locomotion generators in order to realize basic motions of the robots — motion primitives. The proposed planner then constructs the plan using these primitives.

6 citations


Cites background or methods from "A two level fuzzy PRM for manipulat..."

  • ...2.1 Basic path planning methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.2 Randomized motion planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.3 Probabilistic roadmaps (PRM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.3.1 Narrow passage problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.3.2 Workspace-based sampling . . . . . . . . . . . . . . . . . . . . . . . . . ....

    [...]

  • ...In Fuzzy-PRM [185], collision detection is performed only for the nodes during the construction of the roadmap and the edges are not checked....

    [...]

  • ...The idea of the postponed collision detection approach is also part of other PRM-based planners [217, 231, 16, 185, 95]....

    [...]

  • ...Paths for 3D objects are required in assembly/disassembly studies in CAD systems [248, 36, 185, 227, 237, 5]....

    [...]

  • ...in path planning for steerable needles in surgery [274, 6, 182], in assembly/disassembly studies [248, 36, 185, 227, 5] or in computational biology [230, 7, 170, 55, 199, 228]....

    [...]

Dissertation
01 Jan 2009
TL;DR: This thesis derives a multi-objective 4D search algorithm entitled Multi- Step A* (MSA*) based on the seminal A* search algorithm that is proven to find the optimal (least cost) path given a variable successor operator (which enables arbitrary track angle and track velocity resolution), and proposes a self-scheduling replanning architecture for MFP.
Abstract: Unmanned Aerial Vehicles (UAVs) are emerging as an ideal platform for a wide range of civil applications such as disaster monitoring, atmospheric observation and outback delivery. However, the operation of UAVs is currently restricted to specially segregated regions of airspace outside of the National Airspace System (NAS). Mission Flight Planning (MFP) is an integral part of UAV operation that addresses some of the requirements (such as safety and the rules of the air) of integrating UAVs in the NAS. Automated MFP is a key enabler for a number of UAV operating scenarios as it aids in increasing the level of onboard autonomy. For example, onboard MFP is required to ensure continued conformance with the NAS integration requirements when there is an outage in the communications link. MFP is a motion planning task concerned with finding a path between a designated start waypoint and goal waypoint. This path is described with a sequence of 4 Dimensional (4D) waypoints (three spatial and one time dimension) or equivalently with a sequence of trajectory segments (or tracks). It is necessary to consider the time dimension as the UAV operates in a dynamic environment. Existing methods for generic motion planning, UAV motion planning and general vehicle motion planning cannot adequately address the requirements of MFP. The flight plan needs to optimise for multiple decision objectives including mission safety objectives, the rules of the air and mission efficiency objectives. Online (in-flight) replanning capability is needed as the UAV operates in a large, dynamic and uncertain outdoor environment. This thesis derives a multi-objective 4D search algorithm entitled Multi- Step A* (MSA*) based on the seminal A* search algorithm. MSA* is proven to find the optimal (least cost) path given a variable successor operator (which enables arbitrary track angle and track velocity resolution). Furthermore, it is shown to be of comparable complexity to multi-objective, vector neighbourhood based A* (Vector A*, an extension of A*). A variable successor operator enables the imposition of a multi-resolution lattice structure on the search space (which results in fewer search nodes). Unlike cell decomposition based methods, soundness is guaranteed with multi-resolution MSA*. MSA* is demonstrated through Monte Carlo simulations to be computationally efficient. It is shown that multi-resolution, lattice based MSA* finds paths of equivalent cost (less than 0.5% difference) to Vector A* (the benchmark) in a third of the computation time (on average). This is the first contribution of the research. The second contribution is the discovery of the additive consistency property for planning with multiple decision objectives. Additive consistency ensures that the planner is not biased (which results in a suboptimal path) by ensuring that the cost of traversing a track using one step equals that of traversing the same track using multiple steps. MSA* mitigates uncertainty through online replanning, Multi-Criteria Decision Making (MCDM) and tolerance. Each trajectory segment is modeled with a cell sequence that completely encloses the trajectory segment. The tolerance, measured as the minimum distance between the track and cell boundaries, is the third major contribution. Even though MSA* is demonstrated for UAV MFP, it is extensible to other 4D vehicle motion planning applications. Finally, the research proposes a self-scheduling replanning architecture for MFP. This architecture replicates the decision strategies of human experts to meet the time constraints of online replanning. Based on a feedback loop, the proposed architecture switches between fast, near-optimal planning and optimal planning to minimise the need for hold manoeuvres. The derived MFP framework is original and shown, through extensive verification and validation, to satisfy the requirements of UAV MFP. As MFP is an enabling factor for operation of UAVs in the NAS, the presented work is both original and significant.

5 citations

01 Jan 2011
TL;DR: CALM, the combined adaptive locality model, is presented, along with an algorithm to bias path sampling based on the model’s predictions, which delivers a denser sampling of the free path space per unit time than open-loop sampling techniques.
Abstract: Mobile robot motions often originate from an uninformed path-sampling process such as random or low-dispersion sampling. We demonstrate an alternative approach to path sampling that closes the loop on the expensive collision-testing process. Although all necessary information for collision testing a path is known to the planner, that information is typically stored in a relatively unavailable form in a costmap or obstacle map. By summarizing the most salient data in a more accessible form, our process delivers a denser sampling of the free path space per unit time than open-loop sampling techniques. We obtain this result by probabilistically modeling—in real time and with minimal information—the locations of obstacles and free space, based on collision-test results. We present CALM, the combined adaptive locality model, along with an algorithm to bias path sampling based on the model’s predictions. We provide experimental results in simulation for motion planning on mobile robots, demonstrating up to a 330% increase in paths surviving collision test.

5 citations


Cites methods from "A two level fuzzy PRM for manipulat..."

  • ...A greedy approach, called a ‘lazy PRM’ method (Bohlin and Kavraki, 2000; Nielsen and Kavraki, 2000; Song et al., 2001), holds potential for increased performance in uncluttered environments....

    [...]

References
More filters
Journal ArticleDOI
01 Aug 1996
TL;DR: Experimental results show that path planning can be done in a fraction of a second on a contemporary workstation (/spl ap/150 MIPS), after learning for relatively short periods of time (a few dozen seconds).
Abstract: A new motion planning method for robots in static workspaces is presented. This method proceeds in two phases: a learning phase and a query phase. In the learning phase, a probabilistic roadmap is constructed and stored as a graph whose nodes correspond to collision-free configurations and whose edges correspond to feasible paths between these configurations. These paths are computed using a simple and fast local planner. In the query phase, any given start and goal configurations of the robot are connected to two nodes of the roadmap; the roadmap is then searched for a path joining these two nodes. The method is general and easy to implement. It can be applied to virtually any type of holonomic robot. It requires selecting certain parameters (e.g., the duration of the learning phase) whose values depend on the scene, that is the robot and its workspace. But these values turn out to be relatively easy to choose, Increased efficiency can also be achieved by tailoring some components of the method (e.g., the local planner) to the considered robots. In this paper the method is applied to planar articulated robots with many degrees of freedom. Experimental results show that path planning can be done in a fraction of a second on a contemporary workstation (/spl ap/150 MIPS), after learning for relatively short periods of time (a few dozen seconds).

4,977 citations

Proceedings Article
01 Aug 1998
TL;DR: This paper presents a new class of randomized path planning methods, known as Probabilistic Roadmap Methods (prms), which use randomization to construct a graph of representative paths in C-space whose vertices correspond to collision-free con gurations of the robot.
Abstract: Recently, a new class of randomized path planning methods, known as Probabilistic Roadmap Methods (prms) have shown great potential for solving complicated high-dimensional problems. prms use randomization (usually during preprocessing) to construct a graph of representative paths in C-space (a roadmap) whose vertices correspond to collision-free con gurations of the robot and in which two vertices are connected by an edge if a path between the two corresponding con gurations can be found by a local planning method.

533 citations

Proceedings Article
01 Aug 1998
TL;DR: Y@lBEDGF HEIKJ L„DzMPORQSIUT MV@CW2X >A@CBEDGF HeIKJRX\[^]\X,L>z]  ~Uw ’Nw2x  “eUfa3Vx=
Abstract: .0/2143658709;:=A@CBEDGF HEIKJ L DNMPORQSIUT MV@CW2X >Y@CBEDZF HEIUJRX\[^]\X,L=>N] _?`N5a3P/(b-cYd,/21aeZ/Rfa3CN] vpw /Rxzy;{ |6/R:Y54w}_=/R~Kp€(‚wƒ<‚>Y@lBEDGF HEIUJ L„DNMPO QSI…T MV@lWmX >Y@lBEDGF HEIKJRX\[^]†XsL=>N] ‡†/ ˆ‰w2w21‹ŠŒp~‰Ž/ xN3CY@lBEDGF HEIKJ L„DzMPORQSIUT MV@CW2X >A@CBEDGF HEIKJRX\[^]\X,L>z]  ~Uw ‘N’Nw2x  “eUfa3Vx=Y@lBEDGF HEIKJ L„DzMPORQSIUT MV@CW2X >A@CBEDGF HEIKJRX\[^]\X,L>z]

364 citations

Proceedings Article
12 May 1995
TL;DR: This paper addresses the motion planning problem for a robot in presence of movable objects with an overview of a general approach which consists in building a manipulation graph whose connected components characterize the existence of solutions.
Abstract: This paper addresses the motion planning problem for a robot in presence of movable objects. Motion planning in this context appears as a constrained instance of the coordinated motion planning problem for multiple movable bodies. Indeed, a solution path in the configuration space of the robot and all movable objects is a sequence of transit paths where the robot moves alone and transfer paths where a movable object follows the robot. A major problem is to find the set of configurations where the robot has to grasp or release objects. The paper gives an overview of a general approach which consists in building a manipulation graph whose connected components characterize the existence of solutions. Two planners developed at LAAS-CNRS illustrate how the general formulation can be instantiated in specific cases.

175 citations

Proceedings Article
12 May 1995

160 citations