氏名　Chowarit Mitsantisuk
学位の種類　博士（工学）
学位記番号　博甲第561号
学位授与の日付　平成22年8月31日
学位論文題目　High Performance Human-Robot Interaction Using Wire-Based Twin Drive System　（ワイヤベースのツインドライブシステムを用いた高性能人間－ロボット間相互作用）
論文審査委員
　主査　教授　大石　潔
　副査　教授　近藤　正示
　副査　准教授　伊東　淳一
　副査　准教授　宮崎　敏昌
　副査　慶應義塾大学理工学部専任講師　桂　誠一郎

• 論文目次
• 論文要旨
• 論文審査の結果の要旨

［平成22（2010）年度博士論文題名一覧］ ［博士論文題名一覧］に戻る．

Contents
Title Page　p.i
Table of Contents　p.iii
List of Figures　p.vii
List of Tables　p.xi
Acknowledgments　p.xiii
1　Introduction　p.1
　1.1　Background of Robotic Technology　p.1
　1.2　Conventional Robotic Technology　p.5
　1.3　A New Framework of Human-Robot Interaction System　p.8
　1.4　Outline of the Thesis　p.10
　1.5　Note on Block Diagram Used in the Thesis　p.16
2　High Performance of Obsderver Based on Multi-Sensor Integration　p.18
　2.1　Introduction　p.19
　2.2　Sensorless force control with disturbance observer　p.22
　2.3　Sensorless force control with position-acceleration integrated disturbance observer （PAIDO）　p.27
　2.4　Sensorles force contorl with Kalman filter-based disturbance observer with multi-sensor system （KFDOB）　p.29
　2.4.1　Multi-sensor Integration for Accurate Velocities Estimation　p.29
　2.4.2　Implementation of the proposed method　p.32
　2.5　Experiments　p.35
　2.5.1　Experimental Setup　p.35
　2.5.2　Experimental Results　p.36
　2.6　Conclusions　p.45
3　Wire-Based Twin Drive System　p.46
　3.1　Mechanism of Wire-Based Twin Drive System　p.46
　3.1.1　Introduction　p.47
　3.1.2　Mechanical Structure　p.52
　3.1.3　Acceleration Control System Based on Dual Disturbance Observer　p.56
　3.1.4　Experimental Results　p.68
　3.1.5　Conclusion　p.77
　3.2　Identification of Wire-Based Twin Drive System　p.78
　3.2.1　Introduction　p.78
　3.2.2　Parameter identification　p.90
　3.2.3　Results and Discussion　p.94
　3.2.4　Conclusion　p.111
4　Mechanical Stiffness Control Based on Variable Wire rope Tension Control　p.112
　4.1　Relationship Between Velocity Profiles and Mechanical Stiffness　p.112
　4.1.1　Introduction　p.113
　4.1.2　Controller Design　p.117
　4.1.3　Experiments and Results　p.124
　4.1.4　Conclusion　p.132
　4.2　Velocity Control Based on Variable Wire Rope Tensions　p.135
　4.2.1　Introduction　p.135
　4.2.2　Identification of Twin Direct-Drive Motor System with Consideration of Wire Rope Tension　p.138
　4.2.3　Wire Rope Tension Control and Velocity Control　p.138
　4.2.4　Experimental Results　p.143
　4.2.5　Conclusion　p.148
　4.3　Interaction Force Control Based on Variable Wire Rope Tensions　p.149
　4.3.1　Introduction　p.149
　4.3.2　Designing the Controller　p.152
　4.3.3　Experimental Results　p.157
　4.3.4　Conclusion　p.165
5　Human-Robot Interaction System and Its Applications　p.167
　5.1　Smoothing of Interaction Force Generation　p.167
　5.1.1　Introduction　p.168
　5.1.2　Interaction Force Generation Using B-Spline Function　p.171
　5.1.3　Multi-sensor Scheme for Estimation of Human Movements　p.176
　5.1.4　Experimental Results　p.179
　5.1.5　Conclusion　p.186
　5.2　Power Assist Control Based on Human Stiffness Estimation　p.187
　5.2.1　Introduction　p.188
　5.2.2　Modal analysis of disturbance observer　p.191
　5.2.3　Real-time Human Stiffness Estimation and Variable Power Assist control　p.202
　5.2.4　Experimental Results　p.205
　5.2.5　Conclusion　p.210
　5.3　Mechanical Stiffness Control Based on Human Stiffness Estimation　p.212
　5.3.1　Introduction　p.212
　5.3.2　Calculation of Wire Rope Tension Command　p.213
　5.3.3　Experimental Results　p.215
　5.3.4　Conclusion　p.220
6　Conclusions　p.221
　6.1　Discussion　p.221
　6.2　Future Research Topics　p.227
　6.2.1　Bilateral Control　p.227
　6.2.2　Virtual environment with temperature consideration　p.228
　6.2.3　Variable stiffness of the Twin direct-drive motor system　p.229
Bibliography　p.231
List of Achievements　p.240

Industrial machines are used widely today because of their accuracy and high speed. Therefore, most of the mechanisms used for robots are conventional linkage mechanisms that consist of rigid and heavy structures. In order to obtain higher torque, a mechanical reduction-gear is typically applied. However, the main disadvantages of a gear mechanism system are the limitations imposed on system performance due to its nonlinear characteristics, which include backlash, vibration, and nonlinear friction. The rigid mechanical stiffness of such systems is not always suitable for human-robot applications due to the lack of compliant capability and human operator safety. Therefore, it is necessary to design an alternative methodology that guarantees the robustness and contact stability of a force control system. This thesis focuses on the improvement of the new framework of human-robot interaction system. The thesis is organized as follows:
Based on a literature review of the robot technology, the structure of the industrial robot, force estimation method and the controller design were discussed and the novel framework of the human-robot system was proposed as shown in the Chapter 1.
Chapter2 initially introduces the conventional external force estimation by using disturbance observer. The construction of inner loop acceleration control based on disturbance observer for the robustness in motion control system is presented. The outer observer loop is utilized to estimate the external force. The proposed concept of Kalman-filter-based disturbance observer with acceleration sensor is also described and utilized in the sensorless force control system. An acceleration sensor is utilized to enlarge the bandwidth of the force sensation. Chapter3 gives the explanation of the system architecture of the new frame work of human-robot interaction system which is designed by twin direct-drive motor system. To meet the specifications required for human-robot interaction system, we selected the identical direct-drive motors with wire rope mechanism in this application. Since both direct-drive motors have almost the same value of the friction effect and other nonlinearities, it is easy for the proposed system to compensate these deterioration effects. Moreover, this Chapter realizes the modeling and closed-loop identification of the twin direct-drive motor system with variation of the wire rope tension. From the simulation and experimental results, the total stiffness of the twin direct-drive motor system and the bandwidth of mechanical system can be increased by changing wire rope tension. This feature may make the twin direct-drive motor system very useful to provide a safety and smooth interaction force during operation. Chapter4 gives the explanation of the controller design of position and velocity for twin direct-drive motor system. From the identification results of the previous Chapter, the high-tension of wire rope system is obtained a high mechanical bandwidth in the robot systems. The disturbance observer-based eliminates the need for the tension sensors, and may also be used to achieve the precise velocity and position control with the vibration-free performance in the differential mode. In Chapter 4.2, this Chapter also proposes the variable wire rope tension algorithm to change the mechanical bandwidth according to velocity control of the robot. At the high velocity conditions, the velocity results in the differential mode confirm the vibration-free performance of the proposed algorithms. On the other hand, to lengthening the durability of wire rope, the wire rope tension command is reduced during the static motion. In Chapter 4.3, this Chapter also proposes the variable wire rope tension algorithm to change the mechanical bandwidth according to motion movements of the human. Since the friction has the greatest effect at low velocity, the wire rope tension command is chosen to be relatively large at low velocity. On the other hand, the robot must operate more compliant to safe user interaction. Thus, at high velocity, the controller is designed to obtain soft mechanical compliance by decreasing the wire rope tension command.
Chapter5 describes the programmable interaction force for the human movements. To construct the interaction force generation, this research proposes a new method of the interaction force control based on impedance control and B-spline interpolation for the human-robot interaction system. An adjustable force generation profiles are designed and all of the properties of impedance control and B-spline curves can be adopted. This Chapter also describes the variable power assist control method based on a real-time estimation of the stiffness of the human arm. By considering the stiffness in human arm movements, this method increases the efficiency of the force control system and realizes comfortable force for human-robot interaction.
Chapter6 summarizes the validation of the experimental results and some suggestions on extensions and future work are given as well.

　本論文は，「High Performance Human-Robot Interaction Using Wire-Based Twin Drive System（ワイヤベースのツインドライブシステムを用いた高性能人間―ロボット間相互作用）」と題し，6章より構成されている．
　第1章「序論」では，研究背景とロボット技術に関する従来法の概要を示すとともに，本研究の目的および構成を述べている．
　第2章「センサ統合に基づくオブザーバの高性能化」では，まず従来法の外乱オブザーバと位置・加速度統合型外乱オブザーバ（PADIO）について示し，次に提案法カルマンフィルタベース反力外乱オブザーバ（KFDOB）について述べ，力制御の実験結果を示す．従来法の位置情報型外乱オブザーバとPAIDOと提案法オブザーバKFDOBとの比較を実験により行う．実験結果により，提案するKFDOBは従来の外乱オブザーバに比べ広帯域化によるノイズが少なく高剛性の環境への接触に対しても安定であることを示している．
　第3章「可変張力制御による剛性制御」では，まずワイヤベースのツインドライブシステムの構成を示し，提案する制御系について述べている．Hadamard行列で変換された仮想コモンモードのワイヤ張力制御と仮想ディファレンシャルモードの人間の反力制御を独立に実現することができる．ワイヤベースツインドライブシステムの同定するために，2慣性共振モデルを示し，共振と反共振を用いたワイヤベースロボットのパラメータ同定法について述べている．ワイヤ張力変動に対してシステムの共振周波数を変えられることを実験結果から確認している．
第4章「可変張力制御による剛性制御」では，速度制御のための剛性制御，および力制御のための剛性制御の応用について述べている．速度制御系と人間の反力制御系では，速度情報を計測して張力指令値を変化させる．実験結果から，提案法の有効性を確認している．
　第5章「人間―ロボット間相互作用制御の応用」では．提案した反作用力制御を介護のための3つの人間―ロボット間相互作用制御へと応用し，その構成について述べている．はじめに，B-Splineに基づくインピーダンス指令値のスムージングを仮想的且つ任意に設定し，動作中でもリアルタイムに変化させることができることを示している．次に，人間の動作剛性推定に基づき，パワーアシスト比を変化させる方法示し，機械剛性制御変化について述べている．実験結果より，反作用力制御系の設計手法が有効であることを示している．
第6章「結論」では，本論文で得られた研究成果を求め，総括している．
以上のように，本論文では新しいワイヤベースの人間支援型ロボットの気候やセンサ統合に基づくオブザーバの高性能化や新しい人間―ロボット間相互作用制御法を提案している．よって，本論文は工学上及び工業上貢献するところが大きく，博士（工学）の学位論文として十分な価値を有するものと認める．