Running rtabmap on bag of rgb and depth images
Hi,
I want to run ros rtabmap on a bag of rgb and depth images, with the following topics:
rostopic list
/camera/depth/camera_info
/camera/depth/image
/camera/rgb/camera_info
/camera/rgb/image_color
I am using the well known canned bag rgbd_dataset_freiburg1_xyz as a reference example. This bag has the following topics:
rosbag info /mnt/hdd3/avnerm/avner/slam/data/rgbd_dataset_freiburg1_xyz/rgbd_dataset_freiburg1_xyz.bag
...
topics: /camera/depth/camera_info
/camera/depth/image
/camera/rgb/camera_info
/camera/rgb/image_color
/cortex_marker_array
/imu
/tf
In my bag I don't have /imu, /tf topics.
To get the program to flow, I added the following transforms. I added the transform with 0 values for x,y,z,rot,pitch,yaw (i.e. no effect), except for the yaw argument to align the orientation of 3d points with the image
<node pkg="tf" type="static_transform_publisher" name="openni_rgb_optical_frame_to_world" args="0 0 0 0.0 0 -$(arg pi) /openni_rgb_optical_frame /world 100" />
<node pkg="tf" type="static_transform_publisher" name="world_to_kinect" args="0.0 0.0 0.0 0.0 0.0 0.0 /world /kinect 100" />
<node pkg="tf" type="static_transform_publisher" name="world_to_map" args="0.0 0.0 0.0 0.0 0.0 0.0 /world /map 100" />
If I just play my bag, I can see in rviz the images (rgb, depth) and the 3d points (via /voxel_cloud), but in local frame coordinates. With these tf nodes the rtabmap runs, but the /rtabmap/mapData shows incorrect - the 3d points are just laid on top of each other without a proper 3d model.
I created a bag rgbd_dataset_freiburg1_xyz_no_tf.bag without the topics: /cortex_marker_array /imu /tf, (and replaced with static transforms)
With this setting, rtabmap creates the mapData incorrectly as well.
So it looks like the launch file that I am using from here requires the topics that were filtered above.
My questions:
- Can rtabmap handle a bag with just rgb and depth images?
- Is there an example canned bag that only uses rgb and depth images?
Thanks,
Avner