Real-Time Vision-Based Mobile Robot Navigation in Outdoor Environments

Real-Time Vision-Based Mobile Robot Navigation in Outdoor Environments

Bernard Ghanem, Ali Fawaz, Ghassan Karame
"Real-Time Vision-Based Mobile Robot Navigation in Outdoor Environments"
4th annual AUB Student Conference 2005 
Bernard Ghanem, Ali Fawaz, and Ghassan Karame
Mobile Robot, SLAM, Computer Vision, Landmark detection, Localization.
2005
​This paper introduces the design of an outdoor navigation system based on range sensors and camera information intended to perform incremental localization in a previously unknown environment, while being able to navigate successfully from source to destination, thus, avoiding impeding obstacles. Prior studies in the field of Simultaneous Localization and Mapping (SLAM) have shown computer vision (CV) to be an influential detection and recognition tool for indoor settings [2]. However, the application of this technique in dynamic outdoor environments, especially where GPS is no longer efficient, has yet to be conducted and tested. Consequently, our mobile robot system will be capable of localizing itself with respect to predefined identifiable landmarks, which are visually “sensed” and detected. These static landmarks will be used by the robot to provide a better estimate of its location in an attempt to plan a path from its unknown initial position to a chosen destination within its environment, while bypassing dynamic obstructions in its path.