動作デｰタ学習を用いた仮想人間のキｰフレｰムアニメｰション [in Japanese] Keyframe Animations of Virtual Humans via Motion Data Learning [in Japanese]
-
- 向井 智彦 MUKAI Tomohiko
- 豊橋技術科学大学情報工学系 Department of Information and Computer Sciences, Toyohashi University of Technology
-
- 栗山 繁 KURIYAMA Shigeru
- 豊橋技術科学大学情報工学系 Department of Information and Computer Sciences, Toyohashi University of Technology
-
- 金子 豊久 KANEKO Toyohisa
- 豊橋技術科学大学情報工学系 Department of Information and Computer Sciences, Toyohashi University of Technology
Search this Article
Author(s)
-
- 向井 智彦 MUKAI Tomohiko
- 豊橋技術科学大学情報工学系 Department of Information and Computer Sciences, Toyohashi University of Technology
-
- 栗山 繁 KURIYAMA Shigeru
- 豊橋技術科学大学情報工学系 Department of Information and Computer Sciences, Toyohashi University of Technology
-
- 金子 豊久 KANEKO Toyohisa
- 豊橋技術科学大学情報工学系 Department of Information and Computer Sciences, Toyohashi University of Technology
Abstract
代表的な仮想人間の動作生成手法であるキーフレーム法は,各キーフレームにおける姿勢を指定し,その関節回転角を適切な関数で補間することで動作を生成する.しかしすべての姿勢を手作業で設計するには膨大な労力が必要であり,最適な補間関数の設計には熟練を要する.一方,近年ではモーションキャプチャ法を用いて実際の人体運動を計測する手法が主流になっており,計測した動作データの効率的な再利用が重要な課題となっている.本論文では,キーフレーム法に基づく動作生成に動作データの運動学的特徴を効率的に利用する手法を提案する.本手法では,筆者らが既に提案した機械学習による動作生成法を拡張し,動作データを用いて学習の報酬関数を設計することで自然な姿勢を探索する.またキーフレームの補間関数を理論的な人体運動モデルと動作データを用いて設計することで,人間らしい滑らかな動作を生成する.本手法により動作データの自然さを反映した複合的な全身動作を効率的に生成できる.
Journal
-
- The IEICE transactions on information and systems Pt. 2
-
The IEICE transactions on information and systems Pt. 2 88(1), 78-87, 2005-01-01
The Institute of Electronics, Information and Communication Engineers
References: 22
-
1
- The coordination of arm movements : An experimentally confirmed mathematical model
-
FLASH T.
J. Neuroscience 5(7), 1688-1703, 1985
Cited by (2)
-
2
- Formation and control of optimal trajectory in human multi-joint arm movement
-
UNO Y.
Biol. Cybern. 61, 89-101, 1989
Cited by (1)
-
3
- Efficient human motion planning using recursive dynamics and optimal control techniques
-
HUANG G.
ASME Proc. DETC'99, 1999
Cited by (1)
-
4
- Human lift-up motion generation based on identification of time-variant performance index
-
MIYATA N.
Proc. 2002 IEEE/RSJ International Conference on Intelligent Robots and Systems, 2002
Cited by (1)
-
5
- RRT-connect : An efficient approach to single-query path planning
-
KUFFNER J. J.
International Conference on Robotics and Automation, 2000, 2000
Cited by (1)
-
6
- Motion path editing
-
GLEICHER M.
Symposium on Interactive 3D Graphics, 2001, 2001
Cited by (1)
-
7
- Adapting Simulated Behaviors For New Characters
-
HODGINS J. K.
Proc. SIGGRAPH'97, 1997
Cited by (2)
-
8
- Retargeting Motion to New Characters
-
GLEICHER M.
Proc. SIGGRAPH'98, 1998
Cited by (2)
-
9
- Motion texture : A two-level statistical model for character motion synthesis
-
LI Y.
SIGGRAPH 2002, 2002
Cited by (1)
-
10
- Motion capture assisted animation : Texturing and synthesis
-
PULLEN K.
SIGGRAPH 2002, 2002
Cited by (1)
-
11
- Parametrization and range of motion of the ball-and-socket joint
-
BAERLOCHER P.
AVATARS, 2000, 2000
Cited by (1)
-
12
- <no title>
-
SUTTON R. S.
Reinforcement Learning, 1998
Cited by (48)
-
13
- <no title>
-
http://www.vcl.ics.tut.ac.jp/mukai/research/
Cited by (1)
-
14
- Automated extraction and parameterization of motions in large datasets
-
KOVAR L.
SIGGRAPH 2004, 2004
Cited by (1)
-
15
- Complex Character Positioning Based on a Compatible Flow Model of Multiple Supports
-
BOULIC R.
IEEE Transactions on Visualization and Computer Graphics 3(3), 245-261, 1997
DOI Cited by (6)
-
16
- Physically based motion transformation
-
POPOVIC Z.
Proc. SIGGRAPH99, 1999
Cited by (17)
-
17
- Online Motion Retargeting
-
CHOI K. J.
The Journal of Visualization and Computer Animation 11(5), 223, 2000
Cited by (4)
-
18
- Natural Motion Animation through Constraining and Deconstraining at will
-
YAMANE K.
IEEE Transactions on Visualization and Computer Graphics 9(3), 352-360, 2003
Cited by (9)
-
19
- Real-time inverse kinematics techniques for anthropomorphic limbs
-
TOLANI D.
Graphical Models 62, 353-388, 2000
DOI Cited by (7)
-
20
- Transfer of learning by composing solutions of elemental sequential tasks
-
SINGH S. P.
Machine Learning 8, 323-339, 1992
DOI Cited by (19)
-
21
- Motion Generation of Virtual Human with Hierarchical Reinforcement Learning [in Japanese]
-
MUKAI Tomohiko , KURIYAMA Shigeru , KANEKO Toyohisa
The Transactions of the Institute of Electronics,Information and Communication Engineers. 86(6), 886-894, 2003-06-01
References (18) Cited by (1)
-
22
- An Adjustment Method of the Number of States on Q-Learning Segmenting State Space Adaptively [in Japanese]
-
HAMAGAMI Tomoki , KOAKUTSU Seiichi , HIRATA Hironori
The Transactions of the Institute of Electronics,Information and Communication Engineers. 86(7), 490-499, 2003-07-01
References (12) Cited by (3)
Cited by: 2
-
1
- Examination of Support Method for Adding Animation to SCORM Environment [in Japanese]
-
SHIMADA Toshiyuki , TAKAMURA Masayuki , NAKAMURA Naoto
IEICE technical report 105(632), 51-56, 2006-03-04
CiNii PDF - Open Access References (9)
-
2
- Examination on Lifting Motion with Different Start-on Posture, and Study on the Proper Opration using Minimum Jerk Model [in Japanese]
-
MATSUMARU Takafumi , FUKUYAMA Satoshi , SHIMA Kazuyoshi , ITO Tomotaka
Transactions of the Japan Society of Mechanical Engineers C 72(720), 2554-2561, 2006-08-25
CiNii PDF - Open Access References (28) Cited by (1)