Hi Stéphane


1. Having multiple implementations of Kinect ROS nodes is not harmful as
we happily share code. However, to ease the end-user's life I think that
we should agree on a naming convention for node and topic naming. Here
is a first suggestion:
  node name: "kinect"
  point cloud frame: "/kinect"
  point could topic: "/kinect/cloud" (PointCloud + color in "rgb")
  rgb image topic: "/kinect/image" (Image, rgb)
  depth (raw in 8bpp) image topic: "/kinect/depth_image" (Image, mono8)
  accelerometer topic: "/kinect/acc" (Vector3)
  tilt input: "/kinect/tilt" (Int16)
Feel free to discuss it and propose more conventions, for instance for
the params/etc.

While things are rapidly developing and different approaches are being tried out multiple drivers are not harmful and potentially are beneficial as people are exploring how to interface with the device.  Unless there are irreconcilable differences it makes sense in the long run for them to merge.  A few of these differences might be licensing and dependency requirements.  However with the scope of the Kinect device I expect that these can be resolved to everyone's satisfaction. 

I agree that finding a common API for the device will make everyone life easier.  There's also the ccny API documented at http://www.ros.org/wiki/kinect_node  They have laid out a few parameters in addition to a very similar set of topics.  

The most important aspect of making life easy for end-users as well as developers is documentation. 

It's awesome to see the Kinect driver useable so quickly, keep up the great work everyone.
Tully