Küresel görüntüler kullanarak iç mekan reaktif seyrüsefer yönteminin geliştirilmesi

An algorithm for visual topological localization and navigation is proposed. Landmark detection and tracking is performed in successive frames to determine angular changes between repetitive and salient features. This angular state is used to generate heading vector which will lead the robot to the target scene. This reactive navigation behavior is aided with localization and path planing algorithms to complete the SLAM capability. Our algorithm is tested in 2 different natural indoor environments. Results are promising in the sense that the accuracy is enough for autonomous navigation purposes. Room to improve topics are discussed.