Human Head Yaw Estimation Based on Two 3-Axis Accelerometers

The estimation of the orientation an object, and a human head in particular, can be defined by Euler angles: yaw, pitch roll. robust drift-free those angles is usually achieved with data from several sensors such as accelerometers, gyroscopes magnetometers, processed sensor fusion algorithms. However, wearable devices hearing instruments are rarely equipped all sensors, might only have single accelerometer embedded per device. While it possible to retrieve correct roll using accelerometer, estimating yaw more challenging task accelerations around gravity vector cannot detected accelerometer. In context binaural communication spatial hearing, key information that helps achieving dynamic synthesis. This work proposes algorithm aims at two 3-axis accelerometers placed each side head. evaluated acceleration measurements subjects realistic task. results suggest strategy used promising good performance for majority measures long individualized set parameters selected. also current implementation sensitive displacement after calibration or wrong during initialization.