wh200720041/ssl_slam

How can I get the raw depth map and pose?

Closed this issue · 5 comments

cdb0y511 commented

Hi, @wh200720041
I find the ssl_slam use point cloud to do slam.
I wish it could provide poses for depth frame so that it can be used for reconstruction. It is possible to enable depth map and align poses together through ssl_slam?
Is there any difference for RGB align to depth map other vices?

wh200720041 commented

Hi @cdb0y511 and @Zumbalamambo

Thanks for posting question here.

In SSL_SLAM the pose is solved by the point cloud. the pose information can be found at /odom. And the depth image is provided by realsense L515. Maybe you can use them to rebuild the depth map. Unfortunately the depth map is not supported yet because it is mainly used in visual SLAM.

Regards
Han

cdb0y511 commented

Hi, @wh200720041 , thanks for the reply.
I have managed to save point cloud through rosrun pcl_ros pointcloud_to_pcd input:=/map.
However, the typical reconstruction methods need 3 inputs, the RGB frame, the depth frame, and the poses.

  1. Can SSL_SLAM output these three?
  2. In my experience, for l515 align the depth map to RGB frame will introduce extra errors, so I prefer RGB frame to align to the depth frame. But I am not sure if SSL_SLAM utilizes the photometric information to do slam, it will have a big difference between these two settings.
wh200720041 commented

Hi @cdb0y511

  1. The poses can be found in /odom, the rgb and depth images can be found in /camera/color/image_raw /camera/depth/image_raw
  2. Kindly note that SSL_SLAM is a LiDAR SLAM rather than Visual SLAM. the pose is not calculated based on photometric error.
    Thanks
cdb0y511 commented

That's great. I will try it later.
Thanks again.