Throttle message rate for subscribers

asked 2011-09-20 20:06:21 -0600

roehling
1951 ●25 ●28 ●43 http://www.fkie.fraunhofer.de/

I have a publisher with a high data rate (approx. 100 Hz) and multiple subscribers. One of them requires the full data rate, while another is quite happy with one message per second.

My simple solution for the latter is to process every 100th message only, but I was wondering if ROS provides a more elegant solution, such as a message filter or a subscription option which lets me choose the desired message rate.

edit retag flag offensive close merge delete

add a comment

3

answered 2011-09-20 20:36:19 -0600

Martin Günther

11816 ●82 ●159 ●198 https://www.dfki.de/en...

In fact, ROS does. What you're looking for is the drop node from topic_tools.

edit flag offensive delete link

Comments

Thanks for the link. In this particular instance, I would prefer a way to control the rate from within my own node (C++), but I'll keep this package in mind for future reference.

roehling ( 2011-09-20 21:32:05 -0600 )edit

add a comment

2

answered 2011-09-21 02:41:41 -0600

joq
25443 ●181 ●256 ●408

For future reference, REP 106 on polled topics is proposed for Fuerte.

Among other things, that will allow an individual subscriber to request that the publisher transmit messages at a lower frequency.

edit flag offensive delete link

Comments

the REP has been updated to Withdrawn because the proposed functionality will not be implemented for the current ROS client libraries`

lucasw ( 2017-07-30 10:11:33 -0600 )edit

add a comment

0

answered 2011-10-25 07:26:23 -0600

Constantin S
296 ●17 ●19 ●31

Answer/Question. Is there a problem with throttling your subscriber using a sleep statement? Put your subscriber to sleep for 1 second, then poll the topic for the newest message? Or do the callbacks wake sleeping subscribers? Or after sleeping do you get the oldest (as opposed to the newest) message in the queue ... which I guess you could handle by setting the queue to size 1.

edit flag offensive delete link

Comments

I gave this a try in response to #q267413 : https://github.com/lucasw/topic_throttle it looks to be working ok and uses a little less cpu than regular topic_tools throttle (which invokes a callback at the full publisher rate). But maybe it is still a bad approach for some reason?

lucasw ( 2017-07-30 10:15:36 -0600 )edit

add a comment

Throttle message rate for subscribers

3 Answers

Comments

Comments

Comments

Question Tools

Stats

Related questions

Throttle message rate for subscribers edit

3 Answers

Comments

Comments

Comments

Question Tools

Stats

Related questions

Throttle message rate for subscribers