Training in virtual environments via a hybrid dynamic trainer model [Elektronische Ressource] / Hasan Esen

Sujets

• Informatik

Informations

• Publié par: technische_universitat_munchen
• Publié le: 01 janvier 2007
• Nombre de lectures: 14
• Langue: English
• Poids de l'ouvrage: 4 Mo

Extrait

Lehrstuhl fur¨ Steuerungs- und Regelungstechnik
Technische Universit¨at Munc¨ hen
Univ.-Prof. Dr.-Ing./Univ. Tokio Martin Buss
Training in Virtual Environments
via a Hybrid Dynamic Trainer Model
Hasan Esen
Vollst¨ andiger Abdruck der von der Fakult¨ at fur¨ Elektrotechnik und Informationstechnik
der Technischen Universit¨ at Munc¨ hen zur Erlangung des akademischen Grades eines
Doktor-Ingenieurs (Dr.-Ing.)
genehmigten Dissertation.
Vorsitzender: Univ.-Prof. Dr.-Ing. Alexander W. Koch
Prufer¨ der Dissertation:
1. Univ.-Prof. Dr.-Ing./Univ. Tokio Martin Buss
2. Associate Prof. Dr. Ken’ichi Yano, Gifu Univ./Japan
Die Dissertation wurde am 21.06.2007 bei der Technischen Universit¨ at Munc¨ hen einge-
reicht und durch die Fakult¨ at fur¨ Elektrotechnik und Informationstechnik am 17.09.2007
angenommen.Foreword
This thesis is a result of four years of work in the research group of my thesis adviser Prof.
Martin Buss. The seed of the work was planted in Berlin in 2003, while Prof. Ken’ichi
Yano was a guest at the Control Systems Group of Technische Universit¨ at Berlin; it was
full-fledged and completed at the Institute of Automatic Control Engineering, Technische
Universit¨ at Munc¨ hen.
First of all, I would like to thank my Doktorvater Prof. Martin Buss not only for the
excellent research environment, his invaluable scientific advice and encouragement, but
also for giving me the research assistant position which brought me financial comfort to
organize my multi-cultural private life.
This work would not be possible without the fruitful discussions with Prof. Ken’ichi
Yano. My cordial thanks to him for his bright ideas and advice, and for his friendship that
was giving me additional power.
Many thanks to my colleague Dr. Marc Ueberle for his hard and fine work on developing
ViSHaRD devices, which played an important role in this thesis. I am thankful to Angelika
Peer for her help on any hardware or software problems concerning ViSHaRD10. Sincere
thanks to Dr. Stanczyk, for always being there as a friend and also for extending my
knowledge on Polish folk music. Many thanks to Raphaela Groten and Klaas Klasing,
who shared the office with me in the last six months and were very understanding about
my “Dissertation Final Phase” anguish.
To all my students, especially to Andreas Sachsenhauser, Kai B¨ orner, Elvin Eren, Dylan
¨Ilg and Serkan Onder, I thank you very much for the extraordinary work and support.
I feel very lucky to have such a great family. My mother and my sisters were the greatest
power for me, although being thousands of kilometers away. It is hard to find words to
express my gratitude to them. Sizlere her¸sey i¸cin cok¸ cok¸ te¸sekkur¨ ediyorum.Myfatheris
further away than thousands of kilometers, but felt like closest to me during the difficult
moments.
My beloved wife Kasia, to whom I got married twice during my work, deserves an indi-
vidual paragraph. Her proof-reading of my articles and dissertation undoubtedly increased
the quality of my writing. Her joy of living, her never ending energy kept me alive and
optimistic. Dzi¸ ekuj¸ e bardzo Kochanie.
Munich, 2007. Hasan Esenanneme ve babama
...Abstract
This thesis presents a novel virtual reality (VR) training concept that integrates the
trainer or the trainer model into training sessions. As an extension to conventional VR
training systems that rely only on realistic interaction, the students are given the chance to
be corrected by the trainer in a multi-user schema. The trainer is connected to the same
virtual environment as the student via an individual haptic display. As an alternative,
task performing skill of the trainer is captured with hybrid identification methods and
the trainer is replaced with the identified model allowing for a single-user training schema.
Two different identification approaches are successfully applied and presented in this thesis:
The weighted K-means clustering-based method and the stochastically switching dynamics
method. Observations and corrections of the trainer or trainer model are multi-modal, i.e.
can be represented in the form of visual, acoustic and/or haptic signals. The combination
of these possible signals allows for the definition of different training strategies. Enhancing
the training systems with extra features that are not available in a real task is investigated
as well. Two different VR scenarios are developed as test-beds: A bone drilling medical
training system and a push button system. The efficiency of the different training strategies
is checked through a series of user tests. To assess the training results objectively, a metric
depending on the distance between the trainer and student in n dimensional Euclidean
space is introduced and applied. The results validate the efficiency and usability of the
training strategies and hybrid identification methods.
Zusammenfassung
Vorliegende Dissertation stellt ein neuartiges Virtual Reality (VR) Trainingkonzept vor,
das den Trainer oder ein Modell des Trainers in eine Trainingssitzung integriert. Als
Erweiterung zu klassischen VR Trainingssystemen, die nur auf realistischer Interaktion
basieren, haben die Studenten die M¨oglichkeit vom Trainer bzw. vom Trainermodell ko-
rrigiert zu werden. In sogenannten Multi-User Szenarien sind Trainer und Schuler¨ durch
zwei individuelle haptische Displays mit der virtuellen Umgebung verbunden. Die senso-
motorische F¨ ahigkeit des Trainers wird mittels der Methode der hybriden Identifikation
erfasst. In dieser Arbeit werden zwei unterschiedliche Ans¨ atze vorgestellt und erfolgreich
angewendet: Die sogenannte gewichtete K-means Clustering Methode und die stochastisch
umschaltende Dynamik Methode. Nachdem das hybride dynamische Modell des Trainers
entworfen worden ist, besteht die M¨oglichkeit den Trainer durch das Modell zu ersetzen.
Beobachtung und Korrektur durch den Trainer bzw. das Modell erfolgen multimodal,
d.h. in Form von visuellen, akustischen sowie haptischen Signalen. Die Kombination
dieser Signale erm¨ oglicht die Konzeption unterschiedlicher Trainingsstrategien. Desweit-
eren wird die Erweiterung der Trainingssysteme durch zus¨ atzliche Funktionen, die in reellen
Systemen nicht existieren, diskutiert. Zwei VR-Szenarien werden als Testumfeld verwen-
det: Das Bohren von Knochen im Rahmen eines medizinischen Trainingssystems sowie
das Bet¨ atigen virtueller Kn¨ opfe. Die Wirksamkeit der Trainingsstrategien wird durch eine
Reihe von Benutzerstudien untersucht. Um die Ergebnisse objektiv zu bewerten, wird
eine Evaluierungsmethode eingefuhrt,¨ die auf einem Maß im n dimensionalen Euklidischen
Raum basiert, das die Distanz zwischen Trainer und Schuler¨ beschreibt. Die Ergebnisse
best¨atigen die Effizienz und Anwendbarkeit der in dieser Dissertation vorgestellten Train-
ingssysteme.
iii“The mythical story of Icarus and Daedalus is usually related to a warning about flying
too high because the heat of the sun would melt the glue used to hold together the feathers
of the wings - this is probably not a correct interpretation of the warning. The more
probable warning was about the danger of flying too high before adequate training...”
Ray L. PageContents
1 Introduction 1
1.1 ProblemDefinitionandMotivation...................... 1
1.2 StateoftheArt............ 4
1.2.1 MedicalSimulatorswithHaptics ......... 4
1.2.2 VRTrainingSystemswithHaptics........ 6
1.3 MainContributionsandOutlineoftheThesis................ 8
2 Virtual Environments for Training 11
2.1 StateoftheArt....................... 12
2.1.1 Collision Detection ................ 12
2.1.2 HapticRendering....... 13
2.2 GeneralPrinciplesofVRDevelopment.......... 15
2.2.1 Collision Detection and Haptic Rendering .... 15
2.2.2 Bounding Boxes and Binary Search Trees............... 18
2.2.3 PhysicallyDeformableObjects........... 18
2.2.4 Stability of Interaction with Virtual Environments 2
2.3 Multi-UserVirtualEnvironments.. 23
2.4 Bone Drilling Medical Training System .................... 25
2.4.1 Bone Drilling with a Pedal-Like Haptic Display . 25
2.4.2 Bone D with ViSHaRD3 ........... 26
2.4.3 Bone Drilling with V10 27
2.4.4 KinematicsMapping............... 3
2.5 Push-ButonTrainingSystem.... 34
2.6 Summary .......................... 35
3 Trainer Skill as a Hybrid Dynamical System 37
3.1 StateoftheArt:MethodstoModelHumanSkil.............. 37
3.1.1 LinearandNon-linearModelingMethods..... 38
3.1.2 SkilSegmentationandSkilPrimitives...... 38
3.1.3 StochasticMethods................. 39
3.1.4 RepresentingSkilasImpedanceandHybridAutomata....... 39
3.2 HumanDemonstrationDataClustering.................... 40
3.2.1 K-MeansClustering...... 40
3.2.2 WeightedK-Means(WKM)ClusteringinParameterSpace..... 41
3.2.3 OverviewofOtherClusteringMethods...... 43
3.3 HybridIdentificationoftheTrainerSkil................... 4
3.3.1 SkillasSwitchingGeneralizedImpedances.... 44
vContents
3.3.2 LeastSquaresOptimizationforImpedanceModels ......... 45
3.3.3 WKM-BasedApproach..................... 46
3.3.4 StochasticalySwitchingImpedances(SS-Imp)Approach...... 47
3.3.5 ComparisonoftheMethods....... 52
3.4 Stability of the Hybrid Models.... 54
3.5 Applicat