Selecting and commanding groups of robots using a Vision-based natural user interface

This thesis presents a novel method of selecting groups of robots from a population. Using a vision-based gestural user interface a user can select a group by circling robots with his or her hand. A robot equipped with a single camera can determine if it was selected by tracking the user's face and a hand. Tracing the user's hand through the circling of robots forms an ellipse in each robots view. We show that by testing if the face point is within that ellipse the robot can infer if it was within the area circled on the floor. This method is then implemented using four robots and demonstrated as a proof of concept.