| Summary: | 碩士 === 國立暨南國際大學 === 電機工程學系 === 103 === In daily tasks of classification, we often employ human power or robotic arms, and
these systems can only be used in a limited range or they may cause unnecessary
consumption of human resources. Therefore, if we can give robotic arm abilities of
moving and looking, they can play a greater role in the job to achieve the purpose of
object classification and save human power.
This thesis presents an intelligent vehicle with object classification, and lets the
robotic arm overcome the obstacle of distance to execute the pick-and-place action.
The thesis is divided into three main parts: mobile vehicle, robotic arm and image
processing. Combining these three parts to reach the pick-and-place task for object
classification is our major control objective. For this purpose, the computer must
analyze images from camera, and transmit the control instructions to the vehicle for
moving or the robotic arm for gripping the correct object and putting it into the
desired destination.
For the control design of systems, inverse kinematics is used to calculate the
movement of a robotic arm, and the computer analyzes and calculates the coordinate
of objects to manipulate the robotic arm. In the part of image processing, color space
of YCbCr, pixel maximum area algorithm and template matching algorithm are
employed to determine the position and relative distance between object and platform.
The major contribution of this thesis is to integrate the separate systems together for
object classification, even if the vehicle is not in the front of platform or in the
situation that the object is not unique. As long as the system can capture the images of
platform or objects to recognize and grip the object correctly, the purpose of object
classification would be achieved successfully.
|
|---|
