ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question
0

Run Navigation Stack with Maps from Kinect Fusion [closed]

asked 2015-01-12 06:27:44 -0500

kido gravatar image

hello everyone,

i am a really newbie to the ROS system therefore if my question ist stupid or inapropriate please be kindly patient.

I have one Robot (er1) with Kinect as Sensor running under ROS and a Server with Kinect Fusion, which can generate maps. My Question is, is it possible to use Navigation Stack with the maps generated by KinFu? The Tutorial says that Navigation Stack needs either LaserScan or PointCloud as the source for sensor information, but for me its not easy to convert datas from Kinect to LaserScan and PointCloud (and because of performance reasons).

Thanks for answers in advice.

edit retag flag offensive reopen merge delete

Closed for the following reason the question is answered, right answer was accepted by kido
close date 2015-02-05 06:52:34.963273

Comments

What is the output from the KinFu?

David Lu gravatar image David Lu  ( 2015-01-12 21:04:47 -0500 )edit

Hi David Lu, Voxelgrid. Kinfu uses the information from the depth-images generated by Kinect and then reconstruct them to voxelgrid. Sry that i didn't see your comment until today.

kido gravatar image kido  ( 2015-01-15 09:18:55 -0500 )edit

Is that a specific ROS message?

David Lu gravatar image David Lu  ( 2015-01-15 10:26:59 -0500 )edit

Hi David,maps are with datatype nav_msgs::OccupancyGrid, 10241024 Pixels (5 meter 5meter) You can see a Screenshot here, thx.

kido gravatar image kido  ( 2015-01-16 08:56:22 -0500 )edit

Hi! I am also interested in using kinfu for 3D mapping and robot navigation. Please, could you provide some tips on how to integrate kinfu with ROS nav stack? Did you use KinfuLS-ROS-Wrapper from here? Thanks!

Dox gravatar image Dox  ( 2018-04-04 07:25:40 -0500 )edit

1 Answer

Sort by ยป oldest newest most voted
1

answered 2015-01-16 10:15:46 -0500

David Lu gravatar image

The ROS Nav stack can accept OccupancyGrid messages as input. Usually they are input as "static maps" where they cover a large stationary area and are updated infrequently.

I don't know exactly how the Kinect Fusion works, but I could see it theoretically being used as input. However, I don't think it has been tried before.

edit flag offensive delete link more

Comments

Thanks David, i'll take a try then make feedbacks here.

kido gravatar image kido  ( 2015-01-20 06:06:00 -0500 )edit

Ok. Best of luck. Please hit the checkmark to close the question when you can.

David Lu gravatar image David Lu  ( 2015-01-20 09:35:13 -0500 )edit

Thanks again to David. I tried and it works after setting the TF correctly. Neither Laserscan nor PointCloud are sufficient when i use the maps from KinFu as the Source for mapserver.

kido gravatar image kido  ( 2015-02-05 06:51:54 -0500 )edit

Question Tools

3 followers

Stats

Asked: 2015-01-12 06:27:44 -0500

Seen: 306 times

Last updated: Jan 16 '15