I am working on a calibration algorithm where the motion is related to the observability of the parameters that I am calibrating.
Let's say we have a 6 dof trajectory and wants to quantify its motion to see how dynamically it has moved. (e.g, to see if it is covering a wide angle in rpy space.)
For example to quantify the SO(3) we can convert it into so(3) then apply PCA to decompose its covariance to see the scatterness along the principal axises. A widely spread motion will give more observability on the parameter calibration.
Is there any reference paper or document for this problem?