Revision history [back]

Answering your comments, a RGB-D images is simply an rgb image associated with a depth image.

According to the rtabmap documentation, all the infos coming from the kinect are given as entry in the rtabmap node as 3 msgs :

1. A sensor_msgs/Image which gives the rgb image
2. A sensor_msgs/Image which gives the depth image
3. A sensors_msgs/Camerainfo which gives the info of the rgb camera.

So what you can do ss give the left rgb image and left camera info directly as if it was the kinect. Then for the depth image, you add a node which contains a stereo algorithm for exemple this one then you define its output image as the depth image in entry of the rtabmap. If you don't know about it, you should look into what's called "remapping" to do those steps.