search

rvp-group / mdslam

MD-SLAM: Multi-cue Direct SLAM. Implements the first photometric LiDAR SLAM pipeline, that works withouth any explicit geometrical assumption. Universal approach, working independently for RGB-D and LiDAR.
BSD 3-Clause "New" or "Revised" License
146 stars 14 forks source link

export poses for every frame #2

Open xiahongchi opened 1 year ago

xiahongchi commented 1 year ago

Hi, could I export poses for every frame using md_slam?

digiamm commented 1 year ago

Hi, currently this feature is not supported. The ROS node spawns both the poses _keyframe_t (within the origin) and _local_t (within the keyframe), but it is pure odometry, not SLAM. The GraphManager and SLAM, in general, operate only on the _keframe_t poses. If you'd like to do so you need to send to the GraphManager a vector of _local_t for each _keyframe_t. At the end of the computation, when the graph needs to be dumped all the _local_t buffers attached to each _ketframe_t need to be transformed accordingly. Currently, the graph only exports the _keyframe_t.

xiahongchi commented 1 year ago

hi, does ba_md_slam have this feature? I would also like to have a try.

digiamm commented 7 months ago

Sorry for the late reply, I missed it (I don't know if it might be helpful for someone else). Anyway, no ba-mdslam does not support this feature. The only difference within SLAM is that the new package runs on GPU.