| Summary: | 碩士 === 國立臺灣大學 === 機械工程學研究所 === 90 === In this thesis, a visual servoing system has been implemented on soccer robot system. We proposed a method combining RGB (Red, Green, Blue) model and HSI (Hue, Saturation, Intensity) model to define the colors of interest. When searching for blobs, geometry characteristics are used to specify the blobs. A predict method is also used to locate the windows of tracking. In terms of robots'' number identification, different shapes of patterns are adapted as features for recognition. A robust algorithm is used to recognize these shapes, and figures out the correct identification that is important for the strategy system. This method is scaling and orientation invariant.
Except for that, we also proposed a virtual soccer robot platform. This platform provides two functions. One is for remote monitoring. The events happened on real soccer robot can be displayed on computer monitor far away. The other is for strategy development of soccer robot system. This platform can demonstrate the result of the strategy design without real soccer robots. Therefore, the strategy development can be beyond the limitation of space, time or money.
|
|---|
