[Ros-kinect] Difference between openni_node.launch and openni_kinect.launch?

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

[Ros-kinect] Difference between openni_node.launch and openni_kinect.launch?

K Chen
Hi,

I am trying to run the MIT hand&finger detection demos, but the packages' launch file uses openni_kinect.lauch in it, and this launch file no longer exists. So I changed it to openni_node.launch:

<include file="$(find openni_camera)/launch/openni_node.launch"/>

Unfortunately this seems to be different and results in poor performance of skeleton tracking using kinect. For example, user tracking can suddenly lost a target even though he did not change his position, and when doing calibration, it will always lost the target when calibration is nearly done.

I got ros packages recently so I could not find the old openni_kinect.lauch file anywhere, can anyone attach one or tell me the difference? I think the problem may lie in changing of tf transformer because it doesn't have problem accessing depth image.

Thanks in advance.

--
Regards

University of Science and Technology of China
School of Computer Science and Technology

K.Chen

_______________________________________________
Ros-kinect mailing list
[hidden email]
https://code.ros.org/mailman/listinfo/ros-kinect
Reply | Threaded
Open this post in threaded view
|

Re: [Ros-kinect] Difference between openni_node.launch and openni_kinect.launch?

Ugo Cupcic
Hi,

The openni_kinect.launch has been replaced by the openni_node.launch (I think that the only difference is that it now detect that a kinect is plugged in and launch the kinect node on its own).

I'm not sure what you mean by changing of tf transformer: the kinect_frames.launch have not been changed.

Cheers,

Ugo

On Fri, Mar 4, 2011 at 12:51 PM, K Chen <[hidden email]> wrote:
Hi,

I am trying to run the MIT hand&finger detection demos, but the packages' launch file uses openni_kinect.lauch in it, and this launch file no longer exists. So I changed it to openni_node.launch:

<include file="$(find openni_camera)/launch/openni_node.launch"/>

Unfortunately this seems to be different and results in poor performance of skeleton tracking using kinect. For example, user tracking can suddenly lost a target even though he did not change his position, and when doing calibration, it will always lost the target when calibration is nearly done.

I got ros packages recently so I could not find the old openni_kinect.lauch file anywhere, can anyone attach one or tell me the difference? I think the problem may lie in changing of tf transformer because it doesn't have problem accessing depth image.

Thanks in advance.

--
Regards

University of Science and Technology of China
School of Computer Science and Technology

K.Chen

_______________________________________________
Ros-kinect mailing list
[hidden email]
https://code.ros.org/mailman/listinfo/ros-kinect




--
Ugo Cupcic | Shadow Robot Company | [hidden email]
Software Engineer | 251 Liverpool Road |
need a Hand? | London N1 1LX | +44 20 7700 2487
http://www.shadowrobot.com/hand/ @shadowrobot


_______________________________________________
Ros-kinect mailing list
[hidden email]
https://code.ros.org/mailman/listinfo/ros-kinect
Reply | Threaded
Open this post in threaded view
|

Re: [Ros-kinect] Difference between openni_node.launch and openni_kinect.launch?

K Chen
Thanks for your reply

But here comes the problem, when I run skeleton tracker directly(which will fetch kinect sensor msgs inside the node), it goes all well; but when I use openni_camera/openni_node.launch to publish sensor msgs and other nodes(say, hand_detection) to subscribe it to do tracking, then the result is poor.

So has openni_camera node changed the msgs' type that caused this problem?


On Fri, Mar 4, 2011 at 9:05 PM, Ugo Cupcic <[hidden email]> wrote:
Hi,

The openni_kinect.launch has been replaced by the openni_node.launch (I think that the only difference is that it now detect that a kinect is plugged in and launch the kinect node on its own).

I'm not sure what you mean by changing of tf transformer: the kinect_frames.launch have not been changed.

Cheers,

Ugo

On Fri, Mar 4, 2011 at 12:51 PM, K Chen <[hidden email]> wrote:
Hi,

I am trying to run the MIT hand&finger detection demos, but the packages' launch file uses openni_kinect.lauch in it, and this launch file no longer exists. So I changed it to openni_node.launch:

<include file="$(find openni_camera)/launch/openni_node.launch"/>

Unfortunately this seems to be different and results in poor performance of skeleton tracking using kinect. For example, user tracking can suddenly lost a target even though he did not change his position, and when doing calibration, it will always lost the target when calibration is nearly done.

I got ros packages recently so I could not find the old openni_kinect.lauch file anywhere, can anyone attach one or tell me the difference? I think the problem may lie in changing of tf transformer because it doesn't have problem accessing depth image.

Thanks in advance.

--
Regards

University of Science and Technology of China
School of Computer Science and Technology

K.Chen

_______________________________________________
Ros-kinect mailing list
[hidden email]
https://code.ros.org/mailman/listinfo/ros-kinect




--
Ugo Cupcic | Shadow Robot Company | [hidden email]
Software Engineer | 251 Liverpool Road |
need a Hand? | London N1 1LX | +44 20 7700 2487
http://www.shadowrobot.com/hand/ @shadowrobot




--
Regards

University of Science and Technology of China
School of Computer Science and Technology

K.Chen

_______________________________________________
Ros-kinect mailing list
[hidden email]
https://code.ros.org/mailman/listinfo/ros-kinect
Reply | Threaded
Open this post in threaded view
|

Re: [Ros-kinect] Difference between openni_node.launch and openni_kinect.launch?

garratt
I changed the launch files, and the source files to reflect the changes in the openni_camera node, so the topics will work better now.

what do you mean by 'the result is poor'?

Garratt



On Fri, Mar 4, 2011 at 9:07 AM, K Chen <[hidden email]> wrote:
Thanks for your reply

But here comes the problem, when I run skeleton tracker directly(which will fetch kinect sensor msgs inside the node), it goes all well; but when I use openni_camera/openni_node.launch to publish sensor msgs and other nodes(say, hand_detection) to subscribe it to do tracking, then the result is poor.

So has openni_camera node changed the msgs' type that caused this problem?


On Fri, Mar 4, 2011 at 9:05 PM, Ugo Cupcic <[hidden email]> wrote:
Hi,

The openni_kinect.launch has been replaced by the openni_node.launch (I think that the only difference is that it now detect that a kinect is plugged in and launch the kinect node on its own).

I'm not sure what you mean by changing of tf transformer: the kinect_frames.launch have not been changed.

Cheers,

Ugo

On Fri, Mar 4, 2011 at 12:51 PM, K Chen <[hidden email]> wrote:
Hi,

I am trying to run the MIT hand&finger detection demos, but the packages' launch file uses openni_kinect.lauch in it, and this launch file no longer exists. So I changed it to openni_node.launch:

<include file="$(find openni_camera)/launch/openni_node.launch"/>

Unfortunately this seems to be different and results in poor performance of skeleton tracking using kinect. For example, user tracking can suddenly lost a target even though he did not change his position, and when doing calibration, it will always lost the target when calibration is nearly done.

I got ros packages recently so I could not find the old openni_kinect.lauch file anywhere, can anyone attach one or tell me the difference? I think the problem may lie in changing of tf transformer because it doesn't have problem accessing depth image.

Thanks in advance.

--
Regards

University of Science and Technology of China
School of Computer Science and Technology

K.Chen

_______________________________________________
Ros-kinect mailing list
[hidden email]
https://code.ros.org/mailman/listinfo/ros-kinect




--
Ugo Cupcic | Shadow Robot Company | [hidden email]
Software Engineer | 251 Liverpool Road |
need a Hand? | London N1 1LX | +44 20 7700 2487
http://www.shadowrobot.com/hand/ @shadowrobot




--
Regards

University of Science and Technology of China
School of Computer Science and Technology

K.Chen

_______________________________________________
Ros-kinect mailing list
[hidden email]
https://code.ros.org/mailman/listinfo/ros-kinect



_______________________________________________
Ros-kinect mailing list
[hidden email]
https://code.ros.org/mailman/listinfo/ros-kinect
Reply | Threaded
Open this post in threaded view
|

Re: [Ros-kinect] Difference between openni_node.launch and openni_kinect.launch?

K Chen
Hi,
Thanks for doing so.

But where did you pushed your changes? I could not find any change of launch files here:

As I mentioned in my first mail,I tried to run MIT hand detection demo, but the user tracking procedure result is poor because the calibration can never be done(It will always lost the target when calibration is nearly done). And I changed the package's launch file to make it running, where it was "openni_kinect.launch", to "openni_node.launch", that's why I am asking what was the difference.

My running openni or nite user tracking samples respectively are normal and the result comes out well.


On Sat, Mar 5, 2011 at 6:56 AM, garratt gallagher <[hidden email]> wrote:
I changed the launch files, and the source files to reflect the changes in the openni_camera node, so the topics will work better now.

what do you mean by 'the result is poor'?

Garratt




On Fri, Mar 4, 2011 at 9:07 AM, K Chen <[hidden email]> wrote:
Thanks for your reply

But here comes the problem, when I run skeleton tracker directly(which will fetch kinect sensor msgs inside the node), it goes all well; but when I use openni_camera/openni_node.launch to publish sensor msgs and other nodes(say, hand_detection) to subscribe it to do tracking, then the result is poor.

So has openni_camera node changed the msgs' type that caused this problem?


On Fri, Mar 4, 2011 at 9:05 PM, Ugo Cupcic <[hidden email]> wrote:
Hi,

The openni_kinect.launch has been replaced by the openni_node.launch (I think that the only difference is that it now detect that a kinect is plugged in and launch the kinect node on its own).

I'm not sure what you mean by changing of tf transformer: the kinect_frames.launch have not been changed.

Cheers,

Ugo

On Fri, Mar 4, 2011 at 12:51 PM, K Chen <[hidden email]> wrote:
Hi,

I am trying to run the MIT hand&finger detection demos, but the packages' launch file uses openni_kinect.lauch in it, and this launch file no longer exists. So I changed it to openni_node.launch:

<include file="$(find openni_camera)/launch/openni_node.launch"/>

Unfortunately this seems to be different and results in poor performance of skeleton tracking using kinect. For example, user tracking can suddenly lost a target even though he did not change his position, and when doing calibration, it will always lost the target when calibration is nearly done.

I got ros packages recently so I could not find the old openni_kinect.lauch file anywhere, can anyone attach one or tell me the difference? I think the problem may lie in changing of tf transformer because it doesn't have problem accessing depth image.

Thanks in advance.

--
Regards

University of Science and Technology of China
School of Computer Science and Technology

K.Chen

_______________________________________________
Ros-kinect mailing list
[hidden email]
https://code.ros.org/mailman/listinfo/ros-kinect




--
Ugo Cupcic | Shadow Robot Company | [hidden email]
Software Engineer | 251 Liverpool Road |
need a Hand? | London N1 1LX | +44 20 7700 2487
http://www.shadowrobot.com/hand/ @shadowrobot




--
Regards

University of Science and Technology of China
School of Computer Science and Technology

K.Chen

_______________________________________________
Ros-kinect mailing list
[hidden email]
https://code.ros.org/mailman/listinfo/ros-kinect



_______________________________________________
Ros-kinect mailing list
[hidden email]
https://code.ros.org/mailman/listinfo/ros-kinect




--
Regards

University of Science and Technology of China
School of Computer Science and Technology

K.Chen

_______________________________________________
Ros-kinect mailing list
[hidden email]
https://code.ros.org/mailman/listinfo/ros-kinect