It's the latter.
--
You received this message because you are subscribed to the Google Groups "Ceres Solver" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ceres-solver...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ceres-solver/e980077c-a20a-4d83-81dc-8eb7186e7b65%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Thank youAnother query:I am often getteing an error "levenberg_marquardt_strategy.cc:113] Linear solver failure. Failed to compute a finite step."
However it is observed that the given result in such cases is very accurate. For example if I initialize the problem to a point very close to the ground truth then the program is terminating with such message but providing a very good solution.
On the other hand when there is no error message and the program reports CONVERGENCE, often the result is far from the ground truth. Does anyone has an idea why this might be so. Am I doing any mistake. Or is it that Levenberg Marquardt is not the right choice here?
--
You received this message because you are subscribed to the Google Groups "Ceres Solver" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ceres-solver...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ceres-solver/5570553c-947d-45cb-818e-3ac4b5b68d6d%40googlegroups.com.
--
You received this message because you are subscribed to the Google Groups "Ceres Solver" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ceres-solver...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ceres-solver/0b4cbc67-8536-4f80-9b44-de778fa0dec6%40googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ceres-solver/16baa27f-01fc-43db-88a2-4ccec056c2e1%40googlegroups.com.
Holding a rotation constant will of course fix the gauge freedom. In fact, in a different implementation, (with our own optimization method) we keep the rotation of first node equal to zero.However, in the framework of ceres, I found an alternative approach slightly easier to implement. If I fix the rotation of the first node to zero. Then in "ResidualError"I will have to check if any one the two nodes being considered is the first node or not. Depending on that I will have three different types of residual computation. That is also possible. However, instead of that I simply added one more equation/cost function, where I want the rotation of the first node to be equal to zero. These two approaches are equivalent. I just chose the later as it saved a few lines of coding for me. However, I shall try the alternative approach too, to see if anything changes.
To view this discussion on the web visit https://groups.google.com/d/msgid/ceres-solver/3c8c746f-6990-4e85-8063-a143f19f0067%40googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ceres-solver/16baa27f-01fc-43db-88a2-4ccec056c2e1%40googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ceres-solver/CABqdRUBs0vBNthXJ%2BE7_9iSHFcP0qAgME6LL2JZMdGhgDuZrcA%40mail.gmail.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ceres-solver/CAB_h4O7DLPwVmKDF-dAjazg2ohNqC%3DENr3%2B1_MLCuc0WJnoVNA%40mail.gmail.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ceres-solver/CADjmpz9%2BKR8SbO9pqcpP-Nc3Tc5%2BqXCJdj%2BXqn0U6PTmqK9g8w%40mail.gmail.com.
--
You received this message because you are subscribed to the Google Groups "Ceres Solver" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ceres-solver...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ceres-solver/97ccb6d1-fcd3-41d6-a718-7546fa174454%40googlegroups.com.
--
You received this message because you are subscribed to the Google Groups "Ceres Solver" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ceres-solver...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ceres-solver/79192efe-b38a-4aa6-b6d8-9ca81475d853%40googlegroups.com.