Implementation Of a Quaternion-Based Kalman Filter for Human Body Motion Tracking using MARG Sensors

Human body motion tracking using inertial sensors requires an attitude estimation filter capable of tracking in all orientations. One way to represent orientation is to use Euler angles, but they have singularities and therefore are not suitable for human body tracking applications. Quaternions can be used to represent orientation without incurring singu-larities. A quaternion-based Kalman filter has been designed for this purpose and implemented in this thesis. Also, a new suboptimal algorithm to compute the quaternions based on magnetometer and accelerometer data is implemented. This new algorithm called “Factored Quaternion Algorithm” is computationally simpler than previous methods and pro-vides a decoupling property from magnetometer and accelerometer data.