ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question

Revision history [back]

click to hide/show revision 1
initial version

Unfortunately viso2_ros does not support RGBD cameras (yet). The input for the viso2_ros stereo node must be two images from left and right camera.

The underlying library libviso2 would have to be modified a lot to support it. You are very welcome to add this feature!

Unfortunately viso2_ros does not support RGBD cameras (yet). The input for the viso2_ros stereo node must be two images from left and right camera.

The underlying library libviso2 libviso2 would have to be modified a lot to support it. it.

Up to now there exist interfaces for mono odometry and for stereo odometry.

To support RGBD odometry, one would have to create a similar interface for image+depth input. The math would be the same as for stereo odometer but the matcher would have to be extended. You are very welcome to add this feature!

Unfortunately viso2_ros does not support RGBD cameras (yet). The input for the viso2_ros stereo node must be two images from left and right camera.

The underlying library libviso2 would have to be modified a lot to support it.

Up to now there exist interfaces for mono odometry and for stereo odometry.

To support RGBD odometry, one would have to create a similar interface for image+depth input. The math would be the same as for stereo odometer but the matcher would have to be extended. You are very welcome to add this feature!

Another option would be to use the mono odometer of viso2_ros (with all its drawbacks) using just the RGB camera or use other packages that are especially designed for RGBD cameras, e.g. RGBDSLAM.

Unfortunately viso2_ros does not support RGBD cameras (yet). The input for the viso2_ros stereo node must be two images from left and right camera.

The underlying library libviso2 would have to be modified a lot to support it.

Up to now there exist interfaces for mono odometry and for stereo odometry.

To support RGBD odometry, one would have to create a similar interface for image+depth input. The math would be the same as for stereo odometer but the matcher would have to be extended. You are very welcome to add this feature!

Another option would be to use the mono odometer of viso2_ros (with all its drawbacks) using just the RGB camera or use other packages that are especially designed for RGBD cameras, e.g. RGBDSLAM.

EDIT:

If you are looking for an RGBD odometry, fovis was especially designed to work with the Kinect. Unfortunately there is no "official" ROS wrapper for that library (yet), but it should be simple to write one.

Unfortunately viso2_ros does not support RGBD cameras (yet). The input for the viso2_ros stereo node must be two images from left and right camera.

The underlying library libviso2 would have to be modified a lot to support it.

Up to now there exist interfaces for mono odometry and for stereo odometry.

To support RGBD odometry, one would have to create a similar interface for image+depth input. The math would be the same as for stereo odometer but the matcher would have to be extended. You are very welcome to add this feature!

Another option would be to use the mono odometer of viso2_ros (with all its drawbacks) using just the RGB camera or use other packages that are especially designed for RGBD cameras, e.g. RGBDSLAM.

EDIT:

If you are looking for an RGBD odometry, fovis was especially designed to work with the Kinect. Unfortunately there is no "official" ROS wrapper for that library (yet), but it should be simple to write one.

EDIT2: We just released fovis_ros, a wrapper for the above mentioned library. This one works with stereo cameras as well as RGBD cameras such as the Kinect.