RTABmap with iphone depth info, help with tf

[JGittles]

Hey all,

I'm fairly new in my SLAM studies and am working on a project utilizing my iphoneX faceID camera for RTABmapping. at the moment, I have the ability to stream my front facing faceID camera depth, rgb, and camera intrinsic as opencv images and then into sensor_msgs in ros melodic. for reference, my camera_info topic contains primarily empty matrices with the exception of the instrinsic as I am operating under the assumption that the iphone camera API is delivering pre-corrected images.

At the moment, I am trying to implement the kinect diagram in part 2.7 of https://wiki.ros.org/rtabmap_ros/Tutorials/SetupOnYourRobot#Kinect but have had issues with what to use as the TF information to include in the headers of the image messages (at least I assume this is my issue)

my error in the terminal currently reads: odometry: Could not get transform from base_link to (stamp=<timestamp>)... cantransform: source_frame does not exist

sorry for the lack of screenshot, currently on a different computer.

Thank you in advance for the help!

[JGittles]

update. I was able to get everything to work. just had to learn more about tf. Currently trying to tackle visual odomety not finding enough inliers which may or may not be a hardware limit.

[matlabbe]

I would be curious if you can record a small rosbag of the data.