Doing Tagslam in realsense d455f, issue about the bag file.

[Zitong Lan]

Hello everyone, I am tring to implement Tagslam using realsense d455f. I don't need to run the tagslam online. My task is to record a bag file and then do the tagslam to determine the camera location at each timestamp. 

My primary question is about the working flow: should I first do tag detection and then do tagslam algorithm and then visiulize the bag file and get the location and pos of the camera? 

What I can do is to first follow the gettingc started instructions on the website to do tag detection. 

However, I do change the  "image_topics" to the topics I observed in bag file from realsense camera as follow: 

    <rosparam param="image_topics">
      ["/device_0/sensor_1/Color_0/image/data"]
    </rosparam>

but when i run sync_and_detect i can only get the following meesages. what is wrong ?

[Bernd Pfrommer]

You are kind of on the right track.

1) First get intrinsic camera calibration. Either the camera provides it, or you calibrate yourself with a checker board.

2) Collect a rosbag of images with the realsense camera. I don't recall having dealt with color but it may just work.

3) Run sync_and_detect as you did. You need to adjust the ros topics to match what's in the bag. Don't collect compressed images unless you have to. Make sure the launch file for sync_and_detect  has the compression setting right. Also make sure your tag family matches your tag, you are using the right tag detector (MIT can handle double-width tag borders, UMICH not but it's more sensitive), and you have the border width set right (1 vs 2) etc. You can also run the apriltag detector as a separate node and look at the debug images with rqt_image_view to see that the tag detector indeed decodes the tags.

Usually sync_and_detect is showing for every frame how many tags are detected. I see no output, so something is not right.