Errors in derived kinematic variables determined from a fixed accelerometer configuration

Previous authors have described the unstable nature of the differential equations to be solved in order to derive angular velocity components of a rigid body from an accelerometer configuration mounted on the body. Suggestions for alternatives have been made using redundant accelerometers or a combination of accelerometers and gyros. It is the purpose of this paper to present the errors in derived linear and angular kinematic variables consistent with a particular 3-2-1 accelerometer configuration for acceleration experiments conducted at the Naval Biodynamics Laboratory (NBDL). Accelerometer errors of sensitivity, linearity and orientation are considered. The statistics of those errors are obtained from repetitive calibrations of the instrumentation packages and are consistent with the calibration techniques at NBDL. Worst case combinations of the standard deviation errors in sensitivity, linearity and orientation are presented. The performance of the six-accelerometer configuration is compared with the least squares solution using three triaxial accelerometers.