Accelerometer-based Orientation Sensing for Head Tracking in AR & Robotics
This work seeks to improve dynamic accuracy of viewpoint tracking for Augmented Reality. Using an inverted pendulum to model the head, dynamic orientation sensing for a single degree of rotation in a vertical plane is achieved using only a dual axis accelerometer. A unique solution is presented as conventional approaches to solve the model equations fail to produce stable results due to ill conditioning. Accuracy is limited by the noise and model error. However, dynamic tracking with better than 1± accuracy is achieved analytically and experimentally, proving the concept.