Data-driven Covariance Tuning of the Extended Kalman Filter for Visual-based Pose Estimation of the Stewart Platform

This paper explores the quaternion representation in order to devise an extended Kalman filter approach for pose estimation: inertial measurements are fused with visual data so as to estimate the position and orientation of a six degrees-of-freedom rigid body. The filter equations are described along with a data-driven tuning method that selects the model covariance matrix based on experimental results. Finally, the proposed algorithm is applied to a six degrees-of-freedom Stewart platform, a representative system of a large class of industrial manipulators that could benefit from the proposed solution.