The problem of mutual orientation determination of the accelerometer unit relative to body of the motion capture system

Abstract

Two algorithms for determination of the rotation matrix of the inertial unit relative to the motion capture 6DOF target are discussed. Solution of this problem is necessary for further co-operative data processing of these sensors in biomechanical researches. In particular, it is required to estimate the acceleration of a moving object in the field of gravity from the data of a 3D accelerometer and a motion capture system. In this case, the local vertical of the well calibrated motion capture system can be used to separate the gravity induced components from the accelerometer data. Both algorithms use measurements obtained for static positions, which differ in orientation of instrumental axes relative to gravitational vertical and at the same time allow us to estimate systematic errors of accelerometers. The first algorithm assumes orthogonality of instrumental axes of inertial unit and does not take into account errors in information on scale coefficients of accelerometers. The algorithm uses equations wich are linearized in the neighborhood of a priori estimates of the unknown values. The second algorithm does not use a priori information about axes of instrumental frame. In this case, the accelerometer unit is recalibrated, and in addition to the zeros and scale factors, the non-orthogonal angles of the instrumental axes of the unit are determined. The immediate result of its work is a matrix, which allows us to determine projections of apparent acceleration on body axes of the motion capture system. Algorithms are tested on experimental data. The comparison criteria is the repeatability of estimated results obtained from different measurement samples. Processing the results of the trial experiment showed that for inaccurate microelectromechanical systems, the second algorithm produced more stable results.