%PDF-1.6
%
1 0 obj
<>
endobj
2 0 obj
<>stream
2021-06-18T13:43:13+09:00
2021-06-18T13:43:13+09:00
2021-06-18T13:43:13+09:00
Adobe Acrobat 21.0
application/pdf
Ritsumeikan Univ.KO-459
uuid:02f084ba-b089-4d5c-97b2-cd5011451459
uuid:36e915bc-d562-4996-a8c4-f73f1f133f27
Acrobat Web Capture 15.0
endstream
endobj
5 0 obj
<>
endobj
6 0 obj
<>
endobj
3 0 obj
<>
endobj
7 0 obj
<>
endobj
10 0 obj
<>/ProcSet[/PDF/Text]>>/Type/Page>>
endobj
11 0 obj
<>stream
1 g
10 36 574.5 780 re
f
BT
0 g
/TT0 1 Tf
14.25 0 0 14.25 52.6238 788.25 Tm
(Generation of Skilled Multi-joint Reaching Movements under DOF Redundanc\
y)Tj
/TT1 1 Tf
12 0 0 12 467.6758 759.75 Tm
(Masahiro Sekimoto)Tj
-35.265 -2.313 Td
(Human can move dexterously an upper limb through rotation of joints rega\
rdless the degree of freedom of)Tj
-1 -1.5 Td
(the joints is redundant. In such case, however, there exists a myriad of\
arm poses for kinematically)Tj
0 -1.5 TD
(accomplishing a specified endpoint position and they cannot be determine\
d uniquely. This is well-known not)Tj
T*
(only as "Bernstein's DOF problem" in physiology but also as "ill-posedne\
ss of inverse kinematics" in)Tj
T*
(robotics, and leads to the difficulty of motion control of an arm with r\
edundant joints. Human can move such)Tj
T*
(a redundant arm coordinately. If a mechanism of human movements is clari\
fied, robot arms will be)Tj
T*
(manipulated more dexterously.)Tj
1 -2.5 Td
(This paper challenges this unsolved "DOF problem" from physical analysis\
of human reaching movements,)Tj
-1 -1.5 Td
(and approaches to the principle of human skill. It is shown that the coo\
rdinated damping shaping at joints is)Tj
T*
(very important to generate a skilled motion. Differently from the well-k\
nown approaches in motion control of)Tj
T*
(an arm with redundant joints, which derive a unique joint trajectory opt\
imized by some cost function, this)Tj
T*
(paper suggests a novel approach such that a control signal is composed o\
f only a task-space position error)Tj
T*
(feedback together with damping shaping without using precise dynamics pa\
rameters. The effectiveness of the)Tj
T*
(proposed scheme is also discussed.)Tj
1 -2.5 Td
(First, in the case of planar reaching movements by an arm with four join\
ts, an idea named "Virtual spring)Tj
-1 -1.5 Td
(hypothesis" is introduced. The hypothesis translates reaching movements \
as effects of intentional force)Tj
T*
(leading the endpoint to the target expressed as virtual spring-like forc\
e in task space together with joint)Tj
T*
(dampings. In the proposed approach, there remains a possibility that som\
e unexpected self motion is incurred)Tj
T*
(during movements due to the joint redundancy when the joint damping fact\
ors are selected inadequately.)Tj
T*
(However, such a redundancy problem is resolved by determining the joint \
damping factors so as to meet the)Tj
T*
(order of square root of the inertia matrix of the arm. Furthermore, the \
proposed method can generate human-)Tj
T*
(like coordinated motion. The convergence to the target in such a movemen\
t is theoretically verified by using)Tj
T*
(the ideas of "stability on a manifold" and "transferability to the sub-m\
anifold." Next, the virtual spring)Tj
T*
(hypothesis is extended to "Virtual spring-damper hypothesis" by consider\
ing effects of virtual damper in)Tj
T*
(parallel to the virtual spring in task space. It is shown that the addit\
ional virtual damper effects can generate)Tj
T*
(easily quasi-straight endpoint trajectories, which are well-known as typ\
ical characteristics of human skilled)Tj
T*
(movements. The convergence in this case is also verified theoretically. \
These ideas are applied, furthermore,)Tj
T*
(to three-dimensional reaching movements under the gravity effects, motio\
n generations of industrial robot)Tj
T*
(arm with redundant joints, and object manipulations by a hand-arm system\
. Based upon the idea of the virtual)Tj
T*
(spring-damper hypothesis, the task-space iterative learning control \(IL\
C\) is suggested in order to accomplish)Tj
T*
(any trajectory trackings. Its effectiveness is also verified by theoreti\
cal analyses, computer simulations, and)Tj
T*
(experiments. Whereas in the conventional ILC the learning term has been \
constructed almost at the joint)Tj
T*
(level, the proposed method accomplishes trajectory trackings through the\
task-space level learning with)Tj
T*
(adequate damping shapings even in the case of an arm with redundant join\
ts.)Tj
ET
endstream
endobj
12 0 obj
<>
endobj
13 0 obj
(@%;,#kb)
endobj
14 0 obj
<>
endobj
15 0 obj
<>
endobj
18 0 obj
<>
endobj
19 0 obj
<>stream
H\j@zY& [apu 4vH߾*ҜtwO'wSu6cïѕ8&pɄ4]9o{wO?ct|dvulژtEkwELێwˋ-+Y+
B~o[ߌ=5=43,d^d+^zxzxzxv(˙m1ffAf@G#Б@`j
42kl+̎ov`>& f'}q{Fx>6bj?O8vW` '
endstream
endobj
20 0 obj
<>stream
H|y\TG5o^p rfFzFc$Fcv5&*Px%Fos o>/OfDu\5ďf30=j?y]@ |0 zl*x磅&d`LRt
P,ftԩ %%ih3(q\qѱ#u⼶!{X_zऔ_\vȗOZ%&]RzAIbqXtR-B [$
C,nȸmfg;Ԇ C//[(ԗx?
KVHh:)BЫ{'#"atU.Zg %vϕ k
~bF0^ŨѸNI+NSׯ@`P7hhQ&shMnf[l6mm:vs~wig}>}}eULl\C&&
J?Ke])RXy8LASSi)Ϭ1o27+]PM_IQU?5P6TԦ;jNRij:G]nT{}jZWջDŽLJ
4YpeRhqmvuD8>ttp|8pUT?i_SULq:]/K|ShL)LKiSJ*kkir9Z.ϔ2 CtQ\NDCI!!UJ]ŨtU^qLRv*%R<3oL1fs'LY9C^qh}^sRs9N 8
lfu-ee(_]UoN399u>r?xyCܽL̚LQ
TU{x,=7؞
Պ1^k{۽vv.5vڭw7ݻփV)ǬkaVd
н% "[fRKtut_Y$
?geH1/rQ37O&gq| @!LW}={ynxEm^^I
y-]],E;}o|>svO}Yz>||Qs|(kmoo%n싾S*{?rp? XuX
?`g|L#u1a=~Ab%6,QN!qgq%8s(E
08r}1 HB"!1Q)1q-
(d
:A CtP0&lzTR5d:F~M"d"3tNi*3tB)TQCJy*C-jLMpw"tɳ8y6/x!/Z^Kx)/+x%ռײEy\x3o᭼kIwr!ݼ> |rb>'$\g,rZ[uj]2d!3de=dOًy.|/le_J7Gɷ6͏]+t*t8ssle.إ,ņR{Q &&1̐HQi;( 0NdCA"HYJYVw&?g'NtNѩ:Mt678g縜ȗ2_|ͻ=**ګj:W=T<tnd.'p~[OS4gz~W׳=Wz^őr$^1jk^uC^o#"#g""'#e#ޤ7-Uozީw?z8p
rsgrgs7\|...d]v7'b?p 80#p$>#(㢋qssn-?O|op 68<\p1.
\L9o.טK2~48msֹs;NS8NS