Augmented Reality as a Perceptual Aid in Robot Teleoperation

Abstract

Robots have reached a level of sophistication such that it is now productive to include robot-assisted search teams in urban search and rescue scenarios. The perceptual interface to the human operator is frequently a video display that relays images from a camera on the robot. Research into issues surrounding human-robot interaction has suggested that if more intelligence were placed on the robot, it could lessen the load on the human operator and improve performance in the search task. In this thesis we examine the relative performance in a search task where the robot is controlled by a remote human operator under different simulated environmental conditions. Three distinct environmental conditions are represented by: a nominal display from the robot's camera, an artificially degraded display (representing signal interference) of the same information, and an augmented display of the degraded information, in which object recognition algorithms enable the application of a visual overlay to identify a search target. Results show that reduced performance under the degraded condition can be improved to near the nominal level by this augmentation method.