A Mobile Robot Localization using External Surveillance Cameras at Indoor

Abstract Localization is a technique that is needed for the service robot to drive at indoors, and it has been studied in various ways. Most localization techniques let the robot measure environmental information to gain location information, but those require high costs as it use many equipment, and also complicate the robot development. But if an external device could calculate the location of the robot and transmit it to the robot, it will reduce the extra cost for the internal equipment needed to recognize the location, and it will also simplify the robot development. Therefore this study suggests an effective way to control the robot by using the location information of the robot included in a map made by visual information from the surveillance cameras installed at indoors. The object in a single image is difficult to tell its size because of the shadow components and occlusion. Therefore, combination of shadow removal technique using HSV image from indoors and images from different perspective using homography to create two- dimensional map with accurate object information is suggested. In the experiment, the effectiveness of the suggested method is shown by analyzing the movement result of the robot which applied the location information from the two-dimensional map that is based on the multi cameras, which its accuracy is measured in advance.