Research on rotation vector attitude algorithms based on non-uniform sampling

Based on angular incremental signals from practical gyros using non-uniform sampling, the general expressions and error equations for three-sample and two-sample algorithms were presented which revealed the internal links between algorithm drift and sampling time. As a special case of our new algorithm, Miller s algorithm error was found to be closely related to the matrix consisting of sampling points. While equal-interval sampling, on which Miller's theory was based, could make the error matrix singular, altering the sampling intervals could render it full rank, and thereby reduced the algorithm error. However, the non-uniform sampling could increase two-sample algorithm's error. These conclusions were confirmed by simulation and test results.