Adaptive head movements tracking algorithms for AR interface controlled telepresence robot

The paper describes the design of the telepresence robot controlled via the head-mounted display equipped with an inertial measurement unit (IMU). To control the robot and the camera gimbal motion independently with only 3 DOF reference, we developed algorithms based on nonlinear filtering to process IMU measurements and capture head motion patterns. The system is adaptable to different operators' behaviour and works in real-time using only natural head movements. We implement this solution for shared autonomy remotely controlled robots with an augmented reality (AR) interface, which are in a rising demand for technical drills, monitoring and maintenance applications as well as for indoor guided tours.