Dear authors,
Thank you for opening source the great work, CLINS.
Browsing the code, I have a question about the QuadraticIntegralFactor.
First, its usage in the codebase has been commented out, so it seems the factor is not used at all.
Second, according to the comment, its implementation refers to the segmentQuadraticIntegralDiag in Kalibr. However, the kalibr uses a weight matrix Q of size Dim * SplineOrder, whereas CLINS uses a Q of size SplineOrder * SplineOrder. The kalibr implementation apparently follows the derivation in the Sec 4.6 of paper.
Third, to my understanding, the derivation in Kalibr for quadratic integral is for ordinary B-splines on the Euclidean space. That is, it is probably not suitable for cumulative B-splines on Lie groups, esp. SO3 B splines.
So does this QuadraticIntegralFactor in CLINS work in your experience or thinking?
Thanks for your consideration,
Dear authors,
Thank you for opening source the great work, CLINS.
Browsing the code, I have a question about the QuadraticIntegralFactor.
First, its usage in the codebase has been commented out, so it seems the factor is not used at all.
Second, according to the comment, its implementation refers to the segmentQuadraticIntegralDiag in Kalibr. However, the kalibr uses a weight matrix Q of size Dim * SplineOrder, whereas CLINS uses a Q of size SplineOrder * SplineOrder. The kalibr implementation apparently follows the derivation in the Sec 4.6 of paper.
Third, to my understanding, the derivation in Kalibr for quadratic integral is for ordinary B-splines on the Euclidean space. That is, it is probably not suitable for cumulative B-splines on Lie groups, esp. SO3 B splines.
So does this QuadraticIntegralFactor in CLINS work in your experience or thinking?
Thanks for your consideration,