IZUMI Kiyotaka Faculty of Sci.and Eng., Saga University, Research Associate, 理工学部, 助手 (10284613)
SATO Kazuya Graduate School, Saga University, Associate Professor, 大学院・工学系研究科, 助教授 (30284607)
|Budget Amount *help
¥4,000,000 (Direct Cost : ¥4,000,000)
Fiscal Year 2000 : ¥700,000 (Direct Cost : ¥700,000)
Fiscal Year 1999 : ¥1,600,000 (Direct Cost : ¥1,600,000)
Fiscal Year 1998 : ¥1,700,000 (Direct Cost : ¥1,700,000)
In the conventional industrial manipulator, we have to create and incorporate a robot's intelligence part. In tits research, it is an aim that a control system suitable for a robot will be acquired on a simulator only by giving constraints, such as a task command to the robot and collision avoidance, by building the simulator using 3-dimensional simulation and visual sensor techniques without using an expensive actual manipulator. If it is going to directly construct the control-system simulator included to obstacle avoidance of a high degree-of-freedom manipulator by the neural network approach or the fuzzy reasoning technique, a huge network and the control rule which increases exponentially will be needed, and it will become very troublesome work.
This research tries to overcome the trouble by the "fuzzy behavior-based control method" proposed at our laboratory. The concrete result is as follow :
1. Assuming that 6 degree-of-freedom manipulator is to be PUMA type, and dynamics informa
tion is used, it was considered (a) when a direct-control system (that is, torque information on each joint) is acquired ; (b) how to create fine "trajectory generation" for making it take action to a target-position trajectory by the fuzzy behavior-based control approach, while avoiding an obstacle, supposing that a servo control system in the Cartesian-coordinate system has been already incorporated into a robot manipulator system. As a result, it was confirmed that a relevant control system was successfully obtained. Moreover, according to these facts, it was applied to the "trajectory generation" for the position trajectory control accompanied by obstacle avoidance only using a kinematics like a conventional method, and a good result was also obtained similarly.
2. In the 3D animation, a reconstruction method was examined by OpenGL.That is, it was confirmed that the present simulator would be utilized for the communication control of the virtual space animation of the future, because two computers could be communicated and controlled through LAN network by a simplified animation.
3. In the camera servo system for visual tracking, a binocular type with 4 degree-of-freedom was made and a control system based on the kinematics was also developed. Especially, to compensate an effect of the dynamics and of servo gains of the torque command type, it was shown that employing RBF neural network (NN) is useful to reduce the error of tracking.
Thus, the results about the basic version of each component technology, such as the simulator by the behavior-based control approach, the operation and control of animation by LAN, and the development of the prototype of a visual tracking system, were obtained, though overall integration of a system was necessarily unrealizable from time restrictions. Less