An investigation into the accuracy of calculating upper body joint angles using MARG sensors

We investigate magnetic, angular rate, and gravity (MARG) sensor modules for deriving shoulder, elbow and lumbar joint angles of the human body. We use three tri-axial MARG sensors, placed proximal to the wrist and elbow and centrally on the chest, and employ a quaternion-based Unscented Kalman Filter technique to estimate orientations from the sensor data, from which joint angles are calculated based on a simple model of the arm. Tests reveal that the method has the potential to accurately derive specific angles. When compared with a camera-based system, a root mean square difference between 5° - 15° was observed