La lecture en ligne est gratuite
Le téléchargement nécessite un accès à la bibliothèque YouScribe
Tout savoir sur nos offres
Télécharger Lire

These presentee pour obtenir le grade de

De
153 pages
Niveau: Supérieur

  • dissertation


These presentee pour obtenir le grade de Docteur de l'Universite Louis Pasteur Strasbourg I Discipline : Electronique,Electrotechnique, Automatique Specialite : Robotique par Kanako Miura Robot Hand Positioning and Grasping Using Vision Positionnement et saisir par une main robotique a l'aide de la vision Soutenue publiquement le 3 Fevrier 2004 Membres du Jury Directeur de These : M. Michel de Mathelin, Professeur, Universite Louis Pasteur Directeur de These : M. Hikaru Inooka, Professeur, Universite de Tohoku Examinateur : M. Koichiro Deguchi, Professeur, Universite de Tohoku Examinateur : M. Eiji Nakano, Professeur, Universite de Tohoku Invite : M. Jacques Gangloff, Maıtre de conferences, Universite Louis Pasteur Rapporteur Externe : M. Nicolas Chaillet, Professeur, Universite de Franche-comte Rapporteur Externe : M. Koichi Hashimoto, Professeur, Universite de Tokyo Rapporteur Interne : M. Ernest Hirsch, Professeur, Universite Louis Pasteur

  • positioning task using

  • based visual

  • simplex iterative

  • servoing control

  • rapporteur externe

  • docteur de l'universite

  • associate professor

  • professor


Voir plus Voir moins

Vous aimerez aussi

Th`ese pr´esent´ee pour obtenir le grade de
Docteur de l’Universit´eLouisPasteur
Strasbourg I
´ ´Discipline : Electronique,Electrotechnique, Automatique
Sp´ecialit´e : Robotique
parKanako Miura
Robot Hand Positioning and Grasping
Using Vision
Positionnement et saisir par une main robotique
`a l’aide de la vision
Soutenue publiquement le 3 F´evrier 2004
Membres du Jury
Directeur de Th`ese : M. Michel de Mathelin, Professeur, Universit´eLouisPasteur
Directeur de Th`ese : M. Hikaru Inooka, Professeur, Universit´e de Tohoku
Examinateur : M. Koichiro Deguchi, Professeur, Universit´e de Tohoku
Examinateur : M. Eiji Nakano, Professeur, Universit´e de Tohoku
Invit´e: M.Jacques Gangloff,Maˆıtre de conf´erences, Universit´eLouisPasteur
Rapporteur Externe : M. Nicolas Chaillet, Professeur, Universit´e de Franche-comt´e
Rapporteur Externe : M. Koichi Hashimoto, Professeur, Universit´edeTokyo
Rapporteur Interne : M. ErnestHirsch,Professeur,Universit´eLouisPasteurAcknowledgement
I wish to express my greatest gratitude to Professor Hikaru Inooka, Laboratory of
Intelligent Control Systems, Graduate School of Information Sciences, Tohoku Uni-
versity, and Professor Michel de Mathelin, Equipe d’Automatique, Vision et Robo-
tique, Laboratoire des Sciences de l’Image, de l’Informatique et de la T´el´ed´etection,
Strasbourg I University. Without their continuous encouragement, advice, and as-
sistance, I would not have accomplished my doctor thesis.
Sincere thankfulness is also due to Professor Eiji Nakano and Professor Koichiro
Deguchi, Grasuate School of Information Sciences, Tohoku University, for their serv-
ing on the graduate committee and precious comments on this thesis.
I would like to show my appreciation to Professor Nicolas Chaillet, Besan¸ con
University, Associate Professor Koichi Hashimoto, Tokyo University, and Professor
Ernest Hirsch, Strasbourg I University, for their serving as referee on the precedent
thesis committee and precious comments on this thesis.
I would especially like to thank Associate Professor Jacques Gangloff, Strasbourg
I University, and Research Assistant Nobuaki Nakazawa, Gunma University, for
their tireless guidance and assistance on my research work.
I am greatly indebted to Professor Tadashi Ishihara, Fukushima University, Re-
search Assistant Takahiko Ono, and Research Assistant Yasuaki Ohtaki, Tohoku
University. I learned much from them.
I am also grateful to secretary Ms. Noriko Gyoba, Tohoku University, Profes-
sor Joana Carvalho-Ostertag, Strasbourg I University, and fellow students and re-
searchers in LICS and EAVR for their kindness and help.
I also owe Professor Dani`ele Alexandre, former vice-president of Robert Schuman
University, who integrated me into Strasbourg consortium.
My stay in France (from April 2001 to July 2002) was supported by Renault
Foundation, which is greatly acknowledged.Contents
1 General introduction 1
1.1 Positioningtaskusingvision....................... 1
1.2 Graspingtaskapplyinghumanfeatures . ................ 3
1.3 OrganizationofthisDissertation .................... 4
I Positioning 7
2 Existing approaches 9
2.1 Introduction ................................ 9
2.2 Classificationofvisualservoingalgorithms . .............. 9
2.2.1 Cameraconfigurations ...................... 10
2.2.2 Controllevel............................ 11
2.2.3 Feedbackvariables . ....................... 13
2.3 Basicimage-basedvisualservoingcontrollaws . ............ 15
2.3.1 Indirect image-based visual servoing control law . . . . . . . . 15
2.3.2 Direct visual servoing control law . . . . . . . . . 16
2.4 Uncalibratedvisualservoing . ...................... 17
2.4.1 Existingapproaches. 17
2.4.2 Uncalibrated visual servoing using Newton-like methods . . . 19
2.4.3 Simulations ............................ 21
3 Modified simplex method 27
3.1 Introduction ................................ 27
3.2 Fundamentals of Nelder and Mead simplex method . . . . . . . . . . 29
3.2.1 Simplexiterativeprocess ..................... 29
3.2.2 ConstrainedProblems ...................... 32
iii CONTENTS
3.3 Uncalibrated visual servoing task using the simplex method . . . . . 35
3.3.1 Simplexoptimizationprocesswitharobot . .......... 35
3.4 Simulationresults............................. 37
3.4.1 Simulations to compare modified method with original (Nelder
andMead)simplexmethod . .................. 37
3.5 Conclusion. ................................ 45
4 Practical positioning task with simplex algorithm 47
4.1 Introduction 47
4.2 Objectivefunctions . ........................... 49
4.2.1 Illustrative objective function . . . . . . . . . . . . . . . . . . 49
4.2.2 Pixel matching by sum-of-square-difference . . . . . . . . . . . 50
4.2.3 Simulations ............................ 50
4.2.4 Influence of weightings for illustrative objective function . . . 53
4.2.5 The comparison of different cost functions . . . . . . . . . . . 53
4.3 Improvementoftheconvergence ..................... 60
4.3.1 Hybridscheme . ......................... 60
4.3.2 Variablespace .......................... 62
4.4 Solutionforlocalminimum ....................... 66
4.4.1 Switching to Newton-like iterative methods . . . . . . . . . . . 66
4.4.2 Multiplecameras 66
4.4.3 Goingoutofalocalminimum ................. 66
4.5 Experiments with an industrial robot manipulator . . . . . . . . . . . 68
4.6 Conclusion. ................................ 72
II Grasping 73
5 Measurement of human grasping motion 75
5.1 Introduction 75
5.2 displacementofthetargetobject .................... 76
5.2.1 Experimentalset-up ....................... 76
5.2.2 Resultsanddiscussion ...................... 77
5.3 Contactforceinhumangrasping . ................... 82
5.3.1 Experimentalset-up 82
5.3.2 Resultsanddiscussion 82CONTENTS iii
5.4 Motion of human fingertips and their functions in grasping . . . . . . 91
5.4.1 Experimentalset-up ....................... 91
5.4.2 Resultsanddiscussion ...................... 91
5.5 Conclusion. ................................ 99
6 Application to robot hand grasping 101
6.1 Introduction101
6.2 Controlofthefirstcontactforce ....................103
6.2.1 Detection of touch and the first contact force . . . . . . . . . . 103
6.2.2 Reduction of impact force by soft attachments . . . . . . . . . 103
6.2.3 Experiments with a robot hand . . . . . . . . . . . . . . . . . 103
6.3 Positionsynchronization .........................107
6.3.1 Position synchronization to touch a target object . . . . . . . 107
6.3.2 Position and force control . . . . . . . . . . . . . . . . . . . . 108
6.3.3 Simulation. ............................109
6.3.4 Experiments with a robot hand . . . . . . . . . . . . . . . . . 114
6.4 Conclusion. ................................118
7 Conclusion 119
7.1 Conclusionofthisdissertation ......................119
7.2 Specificcontributionsofthisthesis ...................120
7.3 Futureworks ...............................121
A Basic definitions 123
A.1 Coordinatesandpose . .........................123
A.2 Cameraprojectionmodel ........................126
A.3 ImageJacobian ..............................127List of Figures
2.1 Static-eyeconfiguration. ......................... 11
2.2 Hand-eye configuration . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3 Indirectvisualservoing .......................... 12
2.4 Directvisualservoing........................... 12
2.5 Position-based visual servoing . . . . . . . . . . . . . . . . . . . . . . 14
2.6 Image-basedvisualservoing . ...................... 14
2.7 Indirectimage-basedvisualservoing. .................. 16
2.8 Directimage-basedvisualservoing . 16
2.9 Imageatthestartingpose ........................ 21
2.10 Image at the goal pose . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.11 Imageatthestartingpose 22
2.12 Image at the goal pose . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.13 Result of adaptive Gauss-Newton optimization with 2DOF centering
task .................................... 23
2.14 Result of adaptive Gauss-Newton optimization with 6DOF centering
task 24
2.15 Result of adaptive Gauss-Newton optimization with 6DOF . . . . . . 25
3.1 Block diagram with Nelder-Mead simplex method . . . . . . . . . . . 28
3.2 Reflection process of the simplex iteration . . . . . . . . . . . . . . . 30
3.3 Expansion process of the i . . . . . . . . . . . . . . . 30
3.4 Contraction process of the simplex iteration . . . . . . . . . . . . . . 31
3.5 Reduction process of the simplex iteration . . . . . . . . . . . . . . . 31
3.6 FlowchartofNelder-Meadsimplexmethod .............. 34
3.7 Comparison between classical simplex optimization and modified sim-
plexprocesswitharobot. ........................ 36
3.8 Joint angles and their cost functions with 2DOF . . . . . . . . . . . . 38
vvi LIST OF FIGURES
3.9 Trajectory of joint angles with 2DOF . . . . . . . . . . . . . . . . . . 39
3.10 Result of the classical simplex optimization with 6DOF centering task 40
3.11 Result of the modified simplex optimization with 6DOF centering task 41
3.12 Result of classical simplex optimization with 6DOF . . . . . . . . . . 42
3.13 Result of modified simplex with 6DOF . . . . . . . . . . 43
4.1 Starting and goal position of the robot and their images taken with
a camera at the end-effector . . . . . . . . . . . . . . . . . . . . . . . 48
4.2 Concept of the illustrative objective function . . . . . . . . . . . . . . 51
4.3 Convergence of the cost with the complex method. The operation
takes86steps.. .............................. 52
4.4 Illustrative objective function W = 500,W =0 ............ 543 4
4.5e objective function W =0,W =500 553 4
4.6 Illustrative objective function W = 250,W =250........... 563 4
4.7e objective function W = 150,W =450 573 4
4.8 Illustrative objective function . . . . . . . . . . . . . . . . . . . . . . 59
4.9 Sum-of-squared-difference pixel matching . . . . . . . . . . . . . . . . 59
4.10 Combinedobjectivefunction ....................... 60
4.11 Different variables; the variable controlled by simplex operation is
signified with red, by visual feedback with blue . . . . . . . . . . . . . 63
4.12 Targetobject ............................... 64
4.13 Pose error for simplex in Cartesian space . . . . . . . . . . . . . . . . 65
4.14 Pose error for simplex in Joint space . . . . . . . . . . . . . . . . . . 65
4.15 Usingmultiplecameras. ......................... 67
4.16 Starting position and final position of the manipulator and corre-
sponding images acquired by the camera. . . . . . . . . . . . . . . . . 69
4.17 Evolution of the cost function with initial position 1 . . . . . . . . . . 70
4.18 Evolution of the cost with initial positions 2,3, and 4 . . . . 70
4.19 Comparison of the objective functions . . . . . . . . . . . . . . . . . . 71
5.1 Systemarrangementfortheexperimentalset-up ............ 79
5.2 Camera view of experimental object and markers . . . . . . . . . . . 79
5.3 Average of displacement distance of the markers . . . . . . . . . . . . 80
5.4 Trajectory of the markers put on the target object . . . . . . . . . . . 80