As the application area of robots extends from the industrial robots to service robots, there arises a sharp need for intelligent robot agents. Many researches for implementing intelligent robots have been accomplished. In the field of Artificial Intelligence, they define the intelligence of a robot as the behavior planning of a robot based on the representing of knowledge and the deliberative reasoning. On the other hand, there is an approach for the reactive systems that can be explained as a “Sensing and Acting” mechanism using the information of environment.
In this thesis, a behavior-based controller based on the visual information is proposed for implementing an intelligent robot that can be applied to a service robot. In a semi-structured environment as the application area of service robots, a reactive system using local information takes many advantages. In order to apply the behavior-based controller to the robotic arm, one behavior is proposed and fuzzy rule base is applied to fuse the behaviors intelligently.
In addition, a novel camera head that can control the vergence movement of stereo cameras is designed for eye-in-hand configuration. A tendon driving mechanism is applied for a light and compact camera head and reducing backlash. A nonlinear driving mechanism based on the tendon driving mechanism is also proposed to enhance the accuracy of the depth extraction.