New drivers

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

New drivers

Radu B. Rusu
Administrator
As my colleagues mentioned earlier, we are happy to announce complete ROS support for the new OpenNI (www.openni.org)
interface! You can check our progress at www.ros.org/wiki/ni. Feel free to send any questions you might have and join
our growing OpenNI community at [hidden email]!

There's some concerns about whether we'll be going back and forth between our initial ROS kinect efforts and OpenNI. To
clarify this, we're 100% dedicated on working with and improving the OpenNI interface right now, which means all our
future demos and code releases will most likely be based on the new "ni" ROS stack.

Please expect a 0.1 release tomorrow for the new OpenNI stack, with complete support for Kinect, with nodelets, and PCL
launch files -- similar to our previous releases. In the meantime, you can obviously give it a try from our github repo.

Of course, this effort wouldn't be possible without the help of many of our colleagues at Willow Garage, but most
importantly people in the ROS community, who have pioneered the early efforts, and have built impressive demos that are
still virally being watched around the globe. We thank all of you and hope to continue our collaboration in the future!


Cheers,
Radu.
--
http://pointclouds.org
_______________________________________________
[hidden email] / http://pointclouds.org
https://code.ros.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: New drivers

Nicholas Butko
Radu,

I'm a little confused about OpenNI. It seems like a library for "natural interaction", which I take to mean "interpreting rgbd data and getting high level output". But it also includes drivers? Are the drivers separable from the openni library? How is the code organized?

It is disappointing that OpenNI doesn't support OSX out of the box. I looked a bit yesterday at trying to see what is needed to compile OpenNI for OSX, and it doesn't seem like too much -- the code is well factored around operating system specifics, and the linux code is embarrassingly close to just working on OSX. However, the folks at prime sense might have a better idea.  If it was as simple as things like replacing clock_gettime with gettimeofday, I imagine they would have done it already.

Anyway, I wanted to take a stab at getting the kinect up and running on OSX, and I'd like to have an idea about the best way to go about it. Should I focus on openni, or openni_camera, or nite, or some portion or some combination?

Or should I focus more on the kinect package and forget about OpenNI entirely?

--Nick

On Dec 9, 2010, at 2:01 AM, Radu Bogdan Rusu wrote:

> As my colleagues mentioned earlier, we are happy to announce complete ROS support for the new OpenNI (www.openni.org)
> interface! You can check our progress at www.ros.org/wiki/ni. Feel free to send any questions you might have and join
> our growing OpenNI community at [hidden email]!
>
> There's some concerns about whether we'll be going back and forth between our initial ROS kinect efforts and OpenNI. To
> clarify this, we're 100% dedicated on working with and improving the OpenNI interface right now, which means all our
> future demos and code releases will most likely be based on the new "ni" ROS stack.
>
> Please expect a 0.1 release tomorrow for the new OpenNI stack, with complete support for Kinect, with nodelets, and PCL
> launch files -- similar to our previous releases. In the meantime, you can obviously give it a try from our github repo.
>
> Of course, this effort wouldn't be possible without the help of many of our colleagues at Willow Garage, but most
> importantly people in the ROS community, who have pioneered the early efforts, and have built impressive demos that are
> still virally being watched around the globe. We thank all of you and hope to continue our collaboration in the future!
>
>
> Cheers,
> Radu.
> --
> http://pointclouds.org
> _______________________________________________
> [hidden email] / http://pointclouds.org
> https://code.ros.org/mailman/listinfo/pcl-users

_______________________________________________
[hidden email] / http://pointclouds.org
https://code.ros.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: [Ros-kinect] New drivers

Ken Conley
On Thu, Dec 9, 2010 at 7:39 AM, Nicholas Butko <[hidden email]> wrote:
> Radu,
>
> I'm a little confused about OpenNI. It seems like a library for "natural interaction", which I take to mean "interpreting rgbd data and getting high level output". But it also includes drivers? Are the drivers separable from the openni library? How is the code organized?

Engine/Sensor is the driver.  OpenNI is an abstraction of the driver
so that it can work with multiple drivers, plus a plugin model for
providing higher-level capabilities.  NITE is a plugin to the OpenNI
that provides the skeleton/gesture stuff.

> It is disappointing that OpenNI doesn't support OSX out of the box. I looked a bit yesterday at trying to see what is needed to compile OpenNI for OSX, and it doesn't seem like too much -- the code is well factored around operating system specifics, and the linux code is embarrassingly close to just working on OSX. However, the folks at prime sense might have a better idea.  If it was as simple as things like replacing clock_gettime with gettimeofday, I imagine they would have done it already.
>
> Anyway, I wanted to take a stab at getting the kinect up and running on OSX, and I'd like to have an idea about the best way to go about it. Should I focus on openni, or openni_camera, or nite, or some portion or some combination?

I would probably start outside of ROS with the driver (aka ps_engine):

https://github.com/ros-pkg-git/Sensor

As well as OpenNI:

https://github.com/openni

As OpenNI is a plugin model, you'll have to get it compiling first,
then work on the engine.

> Or should I focus more on the kinect package and forget about OpenNI entirely?

Up to you.  At the ROS level, the abstraction is the same, so the base
driver is interchangeable.  The performance of the PS driver is much
better, though, and you get skeleton tracking.  It sounds like there
are others on the OpenNI-dev list you could coordinate with.

 - Ken
_______________________________________________
[hidden email] / http://pointclouds.org
https://code.ros.org/mailman/listinfo/pcl-users