Kimera Semantics not generating 3D mesh with RealSense D455 about kimera-semantics

[sguttikon]

@shabpompeiano Where you able to solve the issue?

[shabpompeiano]

The issue persists also with D435i

[jolags313]

@shabpompeiano Have you solved this issue? I'm experiencing the same thing.

Also, for me, nothing is streaming on the /camera/aligned_depth_to_color/camera_info or /camera/aligned_depth_to_color/image_raw topics. Could you tell me how you got the depth image in the second screenshot you have?

[shabpompeiano]

If you set the argument align_depth:=true when initializing realsense ros, the aligned depth should be published under /camera/aligned_depth_to_color/image_raw.
To get the depth image in rviz, I am showing the topic /camera/depth/image_rect_raw

[jolags313]

Sorry for the late response, but thank you for the advice- I also had to do a lot of troubleshooting with the camera and firmware to get it to work. I now see the same rviz output you do. However, I am still unable to generate any mesh. Have you figured this out?

[shabpompeiano]

I suspect based on the following issues, that the issue is related to the TF transformations.

          I had to resolve all the issues related TF and it worked for me.

Originally posted by @madhu-korada in #75 (comment)

          Hi @julienip,
          The pose is inferred from the TF tree in ROS (it will look for the tfs you mentioned above, `left_cam` or `left_cam_base_link` depending on what you use).
          Nevertheless, you can also subscribe to a pose topic `transform`, which will be used in case you set the parameter `use_tf_transforms`  to false:
          Note that these are parameters already available in voxblox, and therefore the documentation remains the same, note here https://voxblox.readthedocs.io/en/latest/pages/The-Voxblox-Node.html
          use_tf_transforms
          If true the ros tf tree will be used to get the pose of the sensor relative to the world (sensor_frame and world_frame will be used). If false the pose must be given via the transform topic.

Originally posted by @ToniRV in #26 (comment)