Bubblegrams - a human-robot interaction (HRI) technique - uses MR to allow collocated humans and robots to interact directly by visually augmenting their shared physical environment. Bubblegrams uses interactive comic-like graphic balloons that appear above the robot to allow for interaction between humans and robots. A key technical challenge facing Bubblegrams is the detection of the location of the robot within the user's vision; the MR system needs this information to place the bubble. To solve this, we applied a vision algorithm based on Haar-like features to find and track the robot in real time. This paper introduces the Bubblegrams interface and details the vision algorithm used to detect and track the robot.
