Abstract:

This thesis is divided into two parts. One is the modification of a motion planning algorithm based on an extension of the CT-RRTs algorithm, and the other is the robot visual servoing based on motion planning. First, a virtual force field is established to describe the direction and influence region of a static and moving obstacle. When the end-effector finds an obstacle located along its trajectory, the force strength and direction calculated by the virtual force field will lead the robot to a safe space to avoid the obstacle and proceed to the target location safely. In addition, the force strength used for the adjustment of the DD-RRT dynamic range can build a more secure operating environment for robot arms.

The second part of this thesis investigates the robot visual servoing. With the listing of the Wii game console, its somatosensory system quickly became a hot topic; the development of Kinect and its free Software Development Kit (SDK) enable the use of its human skeleton identification and tracking technology to retrieve the skeleton data. Through a conversion of skeleton data, the robot arms can real-time interact with the human like handshake. Finally, this study applies this concept to construct a simple sign language database by imitating human gestures for advanced interaction.