divide calculations between pi and pc

[aman]

Hello,

I have a Realsense D435 and Lidar A2 which is connected to raspberry pi 4 4gb. I am using rtabmap slam for the fusion of two sensors to do localization, mapping, and navigation.

I run launch files of camera, lidar and rtabmap on pi and pi is connect to the client pc to have a visualization on it. I have a problem with pi because it isn't powerful enough to handle calculations it gets crashed all the time.

is there a way to reduce computation on pi? or another recommendation it would be a great help.

thanks

[matlabbe]

What are you using for odometry? I would recommend to use wheel odometry like on this setup.

[aman]

Thanks for a quick reply. I don't have any odometry I am using rtabmap slam for the camera and after a while map was getting messy and the pictures weren't aligned and it is so hard to understand the map.

so I decided to take odometry source from lidar itself to have better map quality but now pi is crashing.

 

[matlabbe]

Indeed, using lidar odometry can be another way. There is an example here with hectorslam or icp_odometry from rtabmap: http://wiki.ros.org/rtabmap_ros/Tutorials/SetupOnYourRobot#Kinect_.2B-_2D_laser

[aman]

I am already using this method. The problem is in this method it does not save the 3d map. It makes 2d map

[matlabbe]

Set explicitly Grid/FromDepth to true if you want the map created from RGB-D camera instead.