24 21 21115025
i 1 1 2 5 2.1.................................. 6 2.1.1........................... 6 2.1.2........................... 7 2.2...................................... 8 2.3............................ 11 2.4.............................. 12 2.5........................... 15 2.6....................... 17 2.6.1........ 17 2.7.......... 20 2.7.1................................ 20 2.7.2..... 21 3 22 3.1.................................. 22 3.2...................................... 23 3.3................................... 24 3.4......................... 26 3.5........................ 28 4 30 4.1................................... 30
ii 4.2.............................. 31 4.3.............. 31 5 34 36 37
1 1 (Virtual Reality: VR) [1][2] [3] VR VR 3 VR 1 1 1
1 2 [4] [5] 2 6 (Inverse Kinematic: IK) [6] 2 Pamplona 1 AR [7] 1 ( ) (CV) [8] ARToolKit 6 [9]
1 3 PC VR/MR/AR 0%
1 4 2 3 4 5
5 2 [8] 6 ARToolKit AR 6 ( 2.1) IP MP 1 CM 2 4 DIP PIP 1 MP 2 4 MP 2 MP 2.1 ( ) DIP PIP MP IP CM 2.1:
2 6 2.1: 2.1 2.1.1 [10] 2.2 IP MP CM( ) CM( ) 0 80 0 60 0 90 0 60 DIP PIP MP( ) MP( ) 0 80 0 100 0 90 25 25 0 80 0 100 0 90 25 25 0 80 0 100 0 90 25 25 0 80 0 100 0 90 25 25 2.2:
2 7 2.1.2 i(i = 1 4) i = 1 i = 4 0 DIP PIP [11]( 2.2) PIP MP ( ) S [12]( 2.3) IP MP ( 2.4) MP CM ( ) S ( 2.5) 2.2: DIP PIP 2.3: PIP MP ( )
2 8 2.4: IP MP 2.5: MP CM ( ) 2.2 2.4 2.2 DIP θ i1 (= f θi1 (θ i2 )) 2.1 2.4 IP θ 01 (= f θ01 (θ 02 )) 2.2 θ i2 i PIP θ 02 MP i MP 2.3 MP ( ) θ i3 i MP ( ) θ i4 i MP ( ) f θi1 (θ i2 ) = 2 3 θ i2 (2.1) f θ01 (θ 02 ) = 4 3 θ 02 (2.2) { (25 0.5 (θ i3 60.0)) θ i4 (25 0.5 (θ i3 60.0)) (θ i3 60 ) 25 θ i4 25 (θ i3 < 60 ) (2.3) 2.2 (Forward Kinematics: FK) 2.1 Denavit-Hartenberg [13] i DH
2 9 2.3 DH 2.4 2.6 2.7 a α d θ a α d θ 1 0 180 0 θ i4 90 2 0 90 0 θ i3 3 L i3 0 0 θ i2 4 L i2 0 0 θ i1 5 L i1 0 0 0 2.3: i DH a α d θ 1 0 90 0 θ 03 2 0 90 0 θ 04 90 3 L 03 90 0 θ 02 4 L 02 0 0 θ 01 5 L 01 0 0 0 2.4: DH θ i1 L i1 θ 01 L 01 θ i2 L i2 θ θ L i4 i3 i3 θ 02 L 02 θ 04 03 L 03 θ 2.6: i DH 2.7: DH 2.3 i F K i (θ i4, θ i3, θ i2, θ i1 ) 2.4 2.4 F K 0 (θ 04, θ 03, θ 02, θ 01 ) 2.5 F K i (θ i4, θ i3, θ i2, θ i1 ) = L i1 sin θ i4 cos(θ i3 + θ i2 + θ i1 ) +L i2 sin θ i4 cos(θ i3 + θ i2 ) + L i3 sin θ i4 cos θ i3 L i1 cos θ i4 cos(θ i3 + θ i2 + θ i1 ) +L i2 cos θ i4 cos(θ i3 + θ i2 ) + L i3 cos θ i4 cos θ i3 L i1 sin(θ i3 + θ i2 + θ i1 ) + L i2 sin(θ i3 + θ i2 ) + L i3 sin θ i3 (2.4)
2 10 F K 0 (θ 04, θ 03, θ 02, θ 01 ) = L 01 cos θ 04 sin θ 03 cos(θ 02 + θ 01 ) L 01 sin θ 04 sin(θ 02 + θ 01 ) +L 02 cos θ 04 sin θ 03 cos θ 02 L 02 sin θ 04 sin θ 02 +L 03 cos θ 04 sin θ 03 L 01 cos θ 03 cos(θ 02 + θ 01 ) + L 02 cos θ 03 cos θ 02 + L 03 cos θ 03 L 01 sin θ 04 sin θ 03 cos(θ 02 + θ 01 ) + L 01 cos θ 04 sin(θ 02 + θ 01 ) +L 02 sin θ 04 sin θ 03 cos θ 02 + L 02 cos θ 04 sin θ 02 +L 03 sin θ 04 sin θ 03 (2.5) DIP θ i1 IP θ 01 2.1 2.2 2.1 2.4 2.2 2.5 2.6 i F K i (θ i4, θ i3, θ i2 ) 2.7 F K 0 (θ 04, θ 03, θ 02 ) F K i (θ i4, θ i3, θ i2 ) = F K 0 (θ 04, θ 03, θ 02 ) = L i1 sin θ i4 cos(θ i3 + 5 3 θ i2) + L i2 sin θ i4 cos(θ i3 + θ i2 ) +L i3 sin θ i4 cos θ i3 L i1 cos θ i4 cos(θ i3 + 5 3 θ i2) + L i2 cos θ i4 cos(θ i3 + θ i2 ) +L i3 cos θ i4 cos θ i3 L i1 sin(θ i3 + 5 3 θ i2) + L i2 sin(θ i3 + θ i2 ) + L i3 sin θ i3 L 01 cos θ 04 sin θ 03 cos 7 3 θ 02 L 01 sin θ 04 sin 7 3 θ 02 +L 02 cos θ 04 sin θ 03 cos θ 02 L 02 sin θ 04 sin θ 02 +L 03 cos θ 04 sin θ 03 L 01 cos θ 03 cos 7 3 θ 02 + L 02 cos θ 03 cos θ 02 + L 03 cos θ 03 L 01 sin θ 04 sin θ 03 cos 7 3 θ 02 + L 01 cos θ 04 sin 7 3 θ 02 +L 02 sin θ 04 sin θ 03 cos θ 02 + L 02 cos θ 04 sin θ 02 +L 03 sin θ 04 sin θ 03 (2.6) (2.7)
2 11 2.3 6 AR 2.8 2.8: CSV LUT CSV LUT (x i, y i ) AR 3 6 3 (x i, y i ) AR- ToolKit ( ) (x i, y i ) L t 2.9 2.8
2 12 X F Z Y L t C Y X ( x i, yi ) 2.9: L t = C + te (2.8) t C e 2.9 F 6 F L t F 2.1 2.4 2.1.2 i f li (θ i3 ) PIP θ i2 (= f θi2 (θ i3 )) 2.3 2.2 MP ( )
2 13 θ i4 MP ( ) f θi2 (θ i3 ) = α i θ i3 3 + β i θ i3 2 + γ i θ i3 (2.9) α i, β i, γ i F K i (θ i4, θ i3, θ i2 ) f θi2 (θ i3 ) i f li (θ i3 ) 2.10 L i1 cos θ 04 cos 5 3 (α iθ 3 i3 + β i θ 2 i3 + γ i θ i3 ) + θ i3 +L i2 cos θ i4 cos(α i θ 3 i3 + β i θ 2 i3 + γ i θ i3 + θ i3 ) + L i3 cos θ i4 cos θ i3 f li (θ i3 ) = L i1 sin θ 04 cos 5 3 (α iθ 3 i3 + β i θ 2 i3 + γ i θ i3 ) + θ i3 +L i2 sin θ i4 cos(α i θ 3 i3 + β i θ 2 i3 + γ i θ i3 + θ i3 ) + L i3 sin θ i4 cos θ (2.10) i3 L i1 sin 5 3 (α iθ 3 i3 + β i θ 2 i3 + γ i θ i3 ) + θ i3 +L i2 sin(α i θ 3 i3 + β i θ 2 i3 + γ i θ i3 + θ i3 ) + L i3 sin θ i3 f l0 (θ 03 ) MP θ 02 (= f θ02 (θ 03 )) 2.5 2.11 CM ( ) θ 04 f θ02 (θ 03 ) = α 0 θ 03 3 + β 0 θ 03 2 + γ 0 θ 03 (2.11) α 0, β 0, γ 0 F K 0 (θ 04, θ 03, θ 02 ) f θ02 (θ 03 ) i f l0 (θ 03 ) 2.12 f l0 (θ 03 ) = L 01 cos θ 04 sin θ 03 cos 7 3 (α 0θ 03 3 + β 0 θ 03 2 + γ 0 θ 03 ) L 01 sin θ 04 sin 7 3 (α 0θ 03 3 + β 0 θ 03 2 + γ 0 θ 03 ) +L 02 cos θ 04 sin θ 03 cos(α 0 θ 03 3 + β 0 θ 03 2 + γ 0 θ 03 ) L 02 sin θ 04 sin(α 0 θ 03 3 + β 0 θ 03 2 + γ 0 θ 03 ) + L 03 cos θ 04 sin θ 03 L 01 cos θ 03 cos 7 3 (α 0θ 03 3 + β 0 θ 03 2 + γ 0 θ 03 ) +L 02 cos θ 03 cos(α 0 θ 03 3 + β 0 θ 03 2 + γ 0 θ 03 ) + L 03 cos θ 03 L 01 sin θ 04 sin θ 03 cos 7 3 (α 0θ 03 3 + β 0 θ 03 2 + γ 0 θ 03 ) +L 01 cos θ 04 sin 7 3 (α 0θ 03 3 + β 0 θ 03 2 + γ 0 θ 03 ) +L 02 sin θ 04 sin θ 03 cos(α 0 θ 03 3 + β 0 θ 03 2 + γ 0 θ 03 ) +L 02 cos θ 04 sin(α 0 θ 03 3 + β 0 θ 03 2 + γ 0 θ 03 ) + L 03 sin θ 04 sin θ 03 (2.12)
2 14 f li (θ i3 ) f l0 (θ 03 ) 2.8 L t L t f li (θ i3 ) f l0 (θ 03 ) L s 2.13 L s = P + sd (2.13) s P d 2.8 L t Q L s R QR QR = R Q 2.8 2.13 QR = (P + sd) (C + te) = P C + sd te (2.14) QR QR e QR d QR e = 0 (2.15) QR d = 0 (2.16) 2.15 2.16 t s t = { d 2 (CP e) (e d)(cp d)} e 2 d 2 (e d) 2 (2.17) s = {(e d)(cp e) e 2 (CP d)} e 2 d 2 (e d) 2 (2.18)
2 15 CP = P C t 2.8 L t Q s 2.13 L s R L s 0.0 s 1.0 s < 0.0 R 1.0 < s R L s R QR L t Q F 2.5 3 Cyclic-Coordinate Descent(CCD) [14] CCD 1 CCD 2.10 ( ) e ( ) g 1. c e E c g G 2. 1. E G 3. c 2. ( ) 4. 5. c b 2. 4. a
2 16 (1) (2) (3) (4) c b g a (7) g c b a e G E g c b a c g a b g c b a e e e e G E (5) (6)(2)~(5)! "#$%&' g c b a E G c g a b e e 2.10: CCD
2 17 6. 2. 5. CCD damping( ) 2.1 2.2 i damping DIP IP PIP θ i2 MP θ 02 DIP θ i1 IP θ 01 2 e G damping 2.6 2.6.1 2.2 2.5 DIP θ i1 PIP θ i2 DIP θ i1 f θi1 (θ i2 ) PIP θ i2j MP ( ) θ i3j f c θ i2 (θ i3j ) f c θ i3 (θ i2j ) j 0 j n j = 0 j = n θ i2j, θ i3j t j f ct θ i2 (t j ) f ct θ i3 (t j ) t n
2 18 MP ( ) θ i4 θ i4 θ i4 t j fθ t i4 (t j ) 2.11 2.12 fθ t i4 (t j ) fθ t i2 (t j ) fθ t i3 (t j ) 2.11: 2.12: F K i (θi4, θi3, θi2) 2.13 2.15
2 19 θ i3 j c ( θ θ i2 j i3 f ) ct θ i 3 f ( t j ) θ i3n t θ i 3 f ( t j ) θ i2 j 2.13: θ i2n θ i3 j θ i2 j c ( θ θ i3 j i2 f ) t θ i 2 f ct θ i 2 f ( t ( t j j ) ) θ i2n 2.14: θ i3n θ i4 j t θi4 f ( t j ) θ i4 n 2.15: θ i4n
2 20 2.7 2.7.1 i k(k i) k k kd, r kd, 2.6 L kd S kd 2.16 r kd L kd kd S kd 2.16: S kd i F K i (θi4, θi3, θi2) F i F i F i
2 21 2.7.2 2.5 2.5:
22 3 3.1 ( ) ( 0% 100% 0% 400%)
3 23 2 200% 100% f θi1 (θ i2 ) f θ01 (θ 02 ) f θi2 (θ i3 ) f θ02 (θ 03 ) 2.6 2.7 i(i=1 4) i(i = 1 4) i 3.2 3.1 ( ( (mm)) 2 ) 1100000000 1670000000 1350000000 1360000000 3.1: 2.3 CSV LUT
3 24 LUT 3.1 3.1: AR 3.3 3.2 3.2 3.3
3 25 ( (mm) (mm)) 140000 200000 165000 177000 3.2: 3.2: ( 260000) 3.3: ( 230000) open 3.4 3.4:
3 26 AR 3.4 0 400 29 3.5 3.13 3.14 3.15 3.5: :400% 3.6: :0% 3.7: :200% 3.8: :300% 3.9: :200% 3.10: :100%
3 27 3.11: :150% 3.12: :100% 3.13: :50% 3.14:
3 28 3.15: 3.14 3.15 29 3.14 3.5 3.15 3.15 131 165 139 175 200
3 29 5
30 4 4.1 CPU : Pentium(R) Dual-Core CPU E5200 2.50GHz PC C++ BUFFALO USB BSW20K04 640 480 pixel( 30fps) 320 240 pixel 640 480 pixel 4.1 4.1:
4 31 4.2 2.6 2.7 4.1 (msec) 14.1 16.3 6.0 0.2 CG 3.3 36.9 4.1: 30fps 74% 10 12fps 4.3 7 1: 4: 7: 3 1: 2:
4 32 3: 4.2 4.4( CG) 4.5 4.7 4.2: 4.3: 4.4: 4.5: 1 4.6: 2
4 33 4.7: 3 1 5.67 2 5.0 3 4.67 4 MP PIP 3.5 CG 1 MP ( ) MP ( ) 4
34 5 0% 400% 3.2 3.3
5 35 MP ( )
36
37 [1] VR Vol.12 No.1 pp.57-65 2007. [2] Vol.11 No.4 pp.469-477 2006. [3] 2012 [4] Vol.J81-D-2 No.1 pp.45-53 1998. [5] IS3-64 2010. [6] (CD-ROM) 2004. [7] Vitor F. Pamplona Leandro A. F. Fernandes Joao Prauchner Luciana P. Nedel e Manuel M. Oliveira The Image-Based Data Glove Proceedings of X Symposium on Virtual Reality (SVR 2008) pp.204-211 2008 [8] Sanshiro Yamamoto Kenji Funahashi Yuji Iwahori A Study for Vision Based Data Glove Considering Hidden Fingertip with Self-Occlusion Proc. SNPD2012 pp 315-320 2012.
38 [9] Billinghurst Mark Vol.4 No.4 pp.607-616 1999. [10] 1989. [11] ELKOURA G and SINGH K Handrix Animating the Human Hand Symposium on Computer Animation - SCA, pp. 110-119, 2003 [12] 2002. [13] S. Hayati K. Tso and G. Roston Robot Geometry Calibration Trans. of IEEE Robotics and Automation Vol.2 pp.947-951 1988. [14] Chris Welman. Inverse kinematics and geometric constraints for articulated figure manipulation M.Sc Thesis Simon Fraser University 1993.