| Summary: | 碩士 === 國立臺灣科技大學 === 資訊工程系 === 95 === Among all sensory data of a remote controlled robot system, visual feedback provides the most straightforward information on constructing spatial presence for users. For the efficiency of controlling remote robots, many visual feedback systems with special wide-angle lenses on cameras were designed to capture massive images. However, these special wide-angle lenses will cause image distortion heavily. Image must be corrected by image processing algorithms before it could be recognized by users.
This thesis focuses on establishing a visual feedback system which provides wide FOV (field-of-view) presence in remote controlled robots. A new optical sys-tem combined with three webcams and two flat mirrors could asynchronously cap-ture multiple images with different FOV from the same POV (position-of-view). These images were then merged and transformed by the PanMultiCamFilter, devel-oped from the DirectShow. By joining a bearing mechanism with a programmable motor controller, the wide FOV visual feedback system could provide more flexible horizontal FOV. Those images captured by the system were transferred to remote user interface (UI) through networks. At the end, a wide FOV visual feedback sys-tem was created with high presence and low image distortion for remote controlled robots.
|
|---|
