While it may be possible for an overhead view to be avaliable, with the ubiquity of security cameras, there may still be hidden objects in a given environments. Furthermore, these objects might occude the important features from the camera's view. This project creates a single mapping scheme to navigate a small wheeled robot, which can be produced at low cost, called the Zumy, through the environment guided by the camera's view. Based on mappings of clutter throughout the scene, the robot attempts to locate hidden objects, defined by AR tags. Under the hood, this project involves several ROS nodes, communicating with both the robot and the server to which the overhead view feeds. The robot communicates camera feeds to locate the object of interest. Notice that control of robot movement over server was a particular limitation for this task.
Please contact Caleb Chuck (caleb_chuck AT yahoo DOT com) for further information.