Research on sampling strategies of UT transformation for space alignment of distributed multi-sensor information

Space alignment is the base of distributed multi-sensor information fusion. UT (Unscented Transformation) has been paid much attention because it is the second-order approximation at least without the requirement of derivable condition in nonlinear transformation. Now, multiple sampling strategies for the UT method has been developed while their comparison and analysis is absent. The comparison is proposed for the symmetric sampling, simplex min skew sampling and simplex sphere sampling strategies in the case of multi-sensor information space alignment from the sphere to Cartesian coordinates. The simulation shows that UT method based on any above strategies is superior to the first-order linearization approximation method while the symmetric sampling strategy is better than the others.