# [Closed] Navigation Stack: Dropped 100.0% of messages so far

Hello everyone

I am having a issue with costmap_2d. When I try to launch a move_base node with costmap_2d using a real Hokuyo sensor it returns me the following message:

[ WARN] [1495115972.915814948]: MessageFilter [target=map laser ]: Dropped 100.00% of messages so far. Please turn the [ros.costmap_2d.message_notifier] rosconsole logger to DEBUG for more information.


Since there is no message_notifier in rosconsole, I turned the message_filter on. Then I got:

[DEBUG] [1495116202.836967251]: MessageFilter [target=map laser ]: Removed oldest message because buffer is full, count now 50 (frame_id=laser, stamp=1495116178.853445)
[DEBUG] [1495116202.837000462]: MessageFilter [target=map laser ]: Added message in frame laser at time 1495116183.833, count now 50


Looking for a way to solve the problem, I found this and this posts. But in my case, the stamp is not the problem, as you can see in the messages above, there is a difference of 5 seconds after 50 messages, which is ok.

In fact, I am able to plot the scan at RVIZ (so, no tf-related problem) and if I try to add an offset of 5 seconds in my scan's stamp, RVIZ says that it should "extrapolate into the future" in order to plot.

Just to add more information, when I ran the exactly same configuration in a simulation, the Navigation Stack works perfectly. Maybe there is a parameter that I am missing between my simulation and the real robot... Idk.

The Final Question: Anyone knows what else than tf transformation and precise stamp costmap_2d needs in order to perform its computation?

Update

So, it turns out it was a problem with time. But not exactly with the stamp on the messages. After I post here, I realized that something was "weird" between two branchs of my tf tree.

Some tf transformations was running directly from my computer and others from the robot. Even if both were running over the same roscore, I had a difference from +- 18 seconds among them.

Then I realized that the time of my computer was not the same from my robot. There was a delay of +- 18 seconds between both computers. And, voila!

I solved my problem putting all tf publishers inside of the robot, however I also found this troubleshooting for syncing between two computers that may work.

I hope it may help someone else. Good luck and be brave!

edit retag close merge delete