Implementation and complexity analysis of orientation estimation algorithms for human body motion tracking using low-cost sensors

Human Body Motion Tracking is becoming one of the key technologies of new generation human-computer interaction. This paper implements human body motion tracking unit based on Inertial Measurement Uints (IMU), which is a combination of a 3-axis magnetometer, gyroscope and accelerometer. In this paper, three sensors' calibration models are set up and two orientation estimation methods with multisensor fusion are implemented based on gradient descent algorithm and complementary filter algorithm respectively. Computational complexity analysis and testing results comparison of two methods are presented. Experimental results show that these two methods have good accuracy compared with the case without multisensor fusion. In the low-power embedded system for human body motion tracking, the complementary filter is more suitable to be adopted in a tracksuit.