teb_local_planner Multithreading

asked 2019-04-11 05:45:05 -0500

jksllk gravatar image


I am currently developing an Ackermann drive robot and I am using teb_local_planner as my local planner. ROS runs on Jetson TX2.

When running the move_base node I get a warning on that the Map update takes from 0.2 to over 1 seconds, while my desired rate is 5Hz.

While running I am monitoring my CPU usage, and I am noticing only ONE of my six cores are really working. One is constantly at 100%, 2-5 is hovering around 50% and the last one is at 0%.

Is there a way to spread the processing more evenly, thus decreasing my loop time?

For information: I have tested different local costmap resolutions and sizes, however I was not able to get the desired resolution nor size to run fast enough. Have a 3x3 and 0.2 as current params, which runs fast enougth, but I want a lower resolution.

Thanks in advance!

edit retag flag offensive close merge delete