Closed Genozen closed 8 months ago
I think a secondary pose graph which keeps tracks of features in each keyframe would be a good start (see ov_secondary as an example). Otherwise, if you could integrate as an initial guess into rtabmap that could be an interesting direction too.
I have updated OpenVINS support for RTAB-Map. You just need to rebuild RTAB-Map with OpenVINS and then set Odom/Strategy to 10. In ROS you can also use OpenVINS through rtabmap_odom instead of using OpenVINS Node directly.
@borongyuan Hi, I am interested to try OpenVINS support for RTAB-Map. Do you have any instruction for that? Could you point me to the relevant repositories?
Could you point me to the relevant repositories?
PRs are already merged. https://github.com/introlab/rtabmap
Hello, I've recently got my d435i to work well with OpenVins and the Odom path is reasonably good.
I would like to start building a map and perform loop-closures and am wondering if you have any suggestions.
I might get a dToF (3D LiDar) so I can have 3D point clouds generated for full mapping generation.
I was thinking of sending the topics to Rtabmap in ROS 2, but this seems like a hurdle in addition to understand the code in rtabmap in addition to OpenVins... (still learning).
Any suggestions is appreciated!
(oh, and I also have an Oak-D pro in hand... but I'll have to tune it and do the same thing to get the VIO to work first...)