| Project/Area Number |
09650298
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Intelligent mechanics/Mechanical systems
|
|---|
| Research Institution | MEIJI UNIVERSITY |
|---|
Principal Investigator |
KOBAYASHI Hiroaki MEIJI UNIVERSITY, PROFESSOR, 理工学部, 教授 (60130811)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
NOBORISAKA Hirokazu MEIJI UNIVERSITY, RESEARCH ASSITANT, 理工学部, 助手
OZAWA Ryuta MEIJI UNIVERSITY, RESEARCH ASSITANT, 理工学部, 助手
HYODO Kazuhito KANAGAWA INSTITUTE OF TECHNOLOGY, ASSOCIATE PROFESSOR, 工学部, 助教授 (10271371)
|
|---|
| Project Period (FY) |
1997 – 2000
|
| Project Status |
Completed (Fiscal Year 2000)
|
| Budget Amount *help |
¥3,700,000 (Direct Cost: ¥3,700,000)
Fiscal Year 2000: ¥1,100,000 (Direct Cost: ¥1,100,000)
Fiscal Year 1999: ¥400,000 (Direct Cost: ¥400,000)
Fiscal Year 1998: ¥500,000 (Direct Cost: ¥500,000)
Fiscal Year 1997: ¥1,700,000 (Direct Cost: ¥1,700,000)
|
|---|
| Keywords | TENDON-DRIVEN / REDUNDANT TENDONS / JOINT STIFFNESS / INPUT-OUTPUT LINEARIZATION / SET-POINT CONTROL / NEURAL NETWORK CONTROL / MULTI AGENT / ROBOT SYSTEM / ロボットハンド / 適応 制御 / 強化学習 / リアプノフ安定性 / 非線形弾性 / ロボット / ゴニオメータ / 筋電位 / 腱駆動ロボットアーム / 冗長性 / 関節剛性調節 / 知的制御 / ニューラルネットワーク制御 |
|---|
| Research Abstract |
This project made a research on robotic arms driven with redundant tendons to develop a robot that could work cooperatively with human workers. The main results consist of three topics, i.e. (1) basic characteristics of tendon-driven robotic mechanisms, (2) controllers for them, and (3) a robot system suitable for them.
For (1), we discussed basic concepts such as controllability with tendons, redundancy of tendon-driven mechanisms, and adjustability of joint stiffness. Then a method to determine suitable joint stiffness for desired tasks, definition of degree of safety for breakage of tendons and a design method of a compensator for the accident were given. Finally, a design method of a tendon-driven finger hand that could adjust the compliance with respect to grasped objects semi-actively was discussed.
For (2), a nonlinear control system that could control the joint angles and the joint stiffness separately was proposed using an exact input-output linearization technique. After that, a set-point position controller and a set-point force controller were investigated to show the robustness and the flexibility for disturbance forces. Neural network controllers were also applied to adapt to the uncertain environment and to adjust the joint stiffness for desired tasks such as positioning and ball hitting tasks.
For (2), we developed a network-based multi-agent robotic system using mobile agents that could be used easily by non-specialists of robot engineering. Functions such as task scheduling, trajectory planning, joint servoing and so on, were realized as individual software agents, and were changed or renewed dynamically through network without stopping the robot. A prototype system was developed for education and evaluated successfully.
|
