ARPushPush: Augmented Reality Game in Indoor Environment

In this paper, we propose ARPushPush, a novel augmented reality game for indoor environment, which uses vision-based tracking and user’s hand gestures. The systems using trackers are too expensive and the calibration between a camera and a tracker, which is indispensable for vision-based interaction, is not easy. The proposed system only uses two cameras for tracking markers and augmenting virtual objects, respectively. The tracking camera detects ARToolKit markers attached on the ceiling and calculates the pose of user’s head on the fly [11]. The HMD camera detects markers attached to the back of user’s hand for interaction and gives augmented view to a user. The transformation matrix between two cameras is updated when markers are visible in each camera. To support wide working area effectively, we propose an algorithm to train multiple markers on the ceiling. In the ARPushPush based on our framework, a user acts to move blocks augmented in a virtual maze to predefined places. The proposed framework allows a user to work in wide area, to interact naturally, and to collaborate with other users.