Host: The Japan Society of Mechanical Engineers
Name : [in Japanese]
Date : May 10, 2017 - May 13, 2017
Remote control robots prevent a risk of secondary accident at the disaster scene. Lots of these robots equips single camera with the body as a vision which sends an image to the operator, however, it is hard to catch the actual depth from the camera image. This study aims to develop an accessibility interface displaying a target grabbing point on the image. The point mean a position where lifts up a target object by the robot arm. We propose a novel processing way to detect an ellipse from an image of a twisted bicycle and we compare our way with a simple Hough translate from the viewpoint of detection accuracy. The result shows that the processing speed of our way is faster than the speed of Hough transform, however, the detection rate extremely declines when the ellipse losses half of points more.
