Kinect Freenect Raspberry Pi3 without RGB?

asked 2018-04-19 19:50:47 -0500

MarksMan0011 gravatar image

I'm hoping to use a Kinect (v1) for navigation using 3d slam algorithms on my Turtlebot 3. It's running using Freenect on the Raspberry Pi 3, but it can be slow and I'm worried it's computationally adding too much to the Pi's workload.

I don't think I need the full RGB data to run the 3d slam algorithms (correct me if that assumption is wrong), so I was wondering if there is a way to not capture/publish that RGB data?

Would this be possible? Would it be useful?

edit retag flag offensive close merge delete

Comments

Hi :) Certainly, it is not proper to use a kenect on raspberry Pi 3. And most image data(include RGB) is published by ImageTransport topic on ROS. so If you don't need the RGB data, do not publish ImageTransport topic.

Gilbert gravatar imageGilbert ( 2018-04-22 20:41:10 -0500 )edit

What do you mean not proper? It is working just a little bit slowly.

I am using Freenect, it does not publish on ImageTransport. I have not seen any similar looking topic. My question is more about whether I do need the RGB data and if whether that would relieve a substantial burden on the pi?

MarksMan0011 gravatar imageMarksMan0011 ( 2018-04-22 22:17:47 -0500 )edit

It is not necessary to 3d slam RGB data. Freenet has several nodes including the RGB camera node. http://wiki.ros.org/freenect_camera The Freenet save all image datas(RGB, depth..) in a buffer and publich image topics, so removing the node will make some improvements.

Gilbert gravatar imageGilbert ( 2018-04-25 02:57:11 -0500 )edit