Improvement of optical tracking-based orientation estimation by fusing gyroscope information

Optical tracking systems (OTS) can provide high position accuracy over a large workspace. However, the orientation from the OTS is related to the distance between markers, which contains large errors when the distance is small or line-of-sight problems occur. The orientation estimation with a gyroscope is complementary to the OTS-based orientation. In this article, an error-state Kalman filter (ESKF) is proposed to fuse them. Two experiments were performed to verify the performance: first, a unit including markers and a gyroscope was placed statically and rotated dynamically in 3-D space. Secondly, two units were used to estimate the relative orientation between the hand and fingers. The static and dynamic orientation errors reduced from 0.39° ± 0.16° and 2.75° ± 1.56° to 0.23° ± 0.02° and 1.50° ± 0.62°, respectively, when the distance between markers was 13 mm. The second experimental results show that the fused method improved the OTS performance by smoothing the estimate, filling the relative orientation during the line of sight period, and correcting the estimation when there were identification problems of markers.