Tracking of Human Joints Using Twist and Exponential Map

Motion tracking system in the home-based environment exhibits attractive advantages for stroke patients. Current methods suffer from incapability of accurately tracking movements with high degree of freedoms. Besides hardly meeting the predefined position during inertial sensor mounting also affects system's performance.

To tackle these challenges, a motion tracking system using twist and exponential map technology is developed in this paper. Firstly, a kinematic model for trunk and upper extremity is designed. Based on this model, twist and exponential map method which updates frames in their initial coordinates instead of transforming coordinates from one frame to another presents high efficiency and convenience in estimating joints' position and orientation. In the experiment, multiple movements are tracked by both of inertial sensor system and optical tracking system. Their comparison verifies this system's high accuracy.