Custom GICP score to the GTSAM for 3D points
130 views
Skip to first unread message
Nhat Nguyen Duy
Aug 8, 2023, 10:24:57 PM8/8/23
to gtsam users
Hello everyone,
Currently, I want to optimize the sensor poses using pose graph optimization (based on GICP scan matching for all frame pairs). When I asked the author regarding the paper I referenced, he replied as follows:
" Yes, the approach you described is correct. Specifically, you need to implement a GICP factor class derived from gtsam::NonlinearFactor that linearizes the GICP error function and returns the linearized function as a gtsam::HessianFactor. The following document would be a help to see how to implement a custom factor for GTSAM.
Currently, I want to optimize the sensor poses using pose graph optimization (based on GICP scan matching for all frame pairs). When I asked the author regarding the paper I referenced, he replied as follows:
" Yes, the approach you described is correct. Specifically, you need to implement a GICP factor class derived from gtsam::NonlinearFactor that linearizes the GICP error function and returns the linearized function as a gtsam::HessianFactor. The following document would be a help to see how to implement a custom factor for GTSAM.
Creating new factor and variable types: https://gtsam.org/doxygen/ "
However, I don't quite understand. Could anyone help me how to customize to GTSAM? Thank you very much
Best regards,
Duy Nhat
Reply all
Reply to author
Forward
0 new messages