The adaptive Kalman filter based on fuzzy logic for inertial motion capture system

• We design a two-step Kalman filter which is suitable for the embedded system.
• We divide the typical Kalman filter into two sub Kalman filters.
• We separately deal with the gravity acceleration vector and geomagnetic vector.
• The outputs of filters are adjusted based on variance matching of fuzzy logic.
• The mathematical model is simplified and the speed of computation is improved.

In the inertial motion capture system, the model complexity and the large amount of computation make the completion of the orientation estimation algorithm rely solely on PC. Because the data processing speed is slow, it is difficult to realize high-speed motion tracking in the embedded system. In order to further expand the application of the motion tracking technology, this paper introduces a two-step Kalman filter, which is suitable for the embedded system. The filter is composed of two sub filters, and is adaptively adjusted based on the variance matching of fuzzy logic. IMU orientation is calculated based on the filtered acceleration vector and the estimated yaw. This approach simplifies the mathematical model, reduces the matrix operations and improves the speed of computation.