Full-pose estimation using inertial and magnetic sensor fusion in structurized magnetic field for hand motion tracking

Abstract

The individual differences and flexible muscular tissue upon bone make it difficult to realize high accuracy hand motion tracking. A full-pose estimation method is proposed by fusing measurements from an Inertial and Magnetic Measurement Unit (IMMU) in structurized magnetic field derived from a marked permanent magnet. Rather than eliminating the non-geomagnetic field, we realize spatial hand motion estimation (including both the location and orientation) by introducing structurized magnetic field to provide increased physical field dimensions without involving field coupling. The IMMU orientation is derived by using the pose matrix expressed by quaternion with complementary filter. Then, 3D position is localized by solving the magnetic inverse problem with derived orientation. Experiments on a developed platform show that the mean individual orientation error of roll, pitch and yaw attitude angle is 0.61°, 0.52° and 0.92°, respectively. In continuous path tracking, the mean localization error and mean orientation error around Z-axis motion and around X-axis motion is (1.56 mm, 0.12°) and (1.2 mm, 0.10°), respectively. The mean full-pose estimation process time is 1.6 ms. This method provides high-precision full-pose estimation for hand gesture recognition.