A Multi-model Human Motion Tracking Approach with Wearable IMU Sensors

In this paper, we propose a multi-model human motion tracking approach based on wearable IMU sensors to track the user’s movements in real time. According to the four states of feet in the daily activities of human body, four kinematic models are established, namely the left-foot-root (LFR) kinematics model, the right-foot-root (RFR) kinematics model, and the double-foot-root (DFR) kinematics model (the above three are collectively referred to as foot-root kinematics models) and the pelvis-root (PR) kinematics model. And then a model switching mechanism is established. These kinematics models can dynamically switch during tracking. In the foot-root kinematic model, the displacement of the human body can be obtained according to the attitude transformation of the leg. In the pelvis-root kinematic model, the quadratic integration of the acceleration of the sensor accelerometer output is used to calculate the displacement. The four models complement each other so that the human motion can be tracked well. Meanwhile this multimodel human motion tracking approach is versatile and provides greater flexibility in capturing outdoor sports.