-
Notifications
You must be signed in to change notification settings - Fork 98
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Sporadically missing base_link tf frame for Marble HD2 robot #978
Comments
It appears to be the "Bad robot" solution container
"Good robot" solution container:
|
Couple more points while this is still running:
I will need to stop this shortly since it is running in aws, but it appears that either the |
I see the same behavior all the time sporadically with all kinds of topics. A good example is MARV robot with the 4 flippers, which are instructed to lift up in the beginning. Sometimes, only 3 or 2 of them lift up and the rest of them is "dead" for the whole simulation. I guess this behavior has something to do with the amount of nodes being started up at once - there's probably some bug in either ign-transport or ROS... I was reproducing this quite reliably with 3 MARV robots - the 3rd one never moved... But I haven't tried this config for over a month, so maybe something has changed... |
I'm pretty sure there's no time to correctly solve this issue. But could we ask DARPA for a workaround? Would you check after each simulation whether each robot has moved at least a tiny bit, and if there is one that hasn't, restart the run? It'd be really a pity to lose points just because ros-ign publishers not publishing... |
I agree, that sounds like a good workaround. It also should be possible to detect a case like this from the rostopic stats logs, which I believe includes ros topic frequency. Maybe something could automatically detect if one of the expected topics is not being published |
We also observe situations where robot is not starting (robotika, |
Probably another example |
I believe I've been seeing this same issue periodically too, but haven't been taking the time to debug what is causing it until I saw it locally (well accessible to me on aws anyway). It does seem to be affect a variety of topics, actually also noticed in one run the other day a UAV was missing a depth camera topic. For the missing pose topic, I didn't take the time to carefully debug the affected
|
We are seeing this same type of issue on cloudsim sometimes. In some of the run logs from cloudsim, we have vehicles do a tf lookup and the vehicles frames don't exist at all. For example, I launched 6 vehicles, and one of them had this issue and was unable to initialize properly. Throughout the log file, I get the following error: 2217.564000000 WARN /A2/cartographer_node [/tmp/binarydeb/ros-melodic-cartographer-ros-1.0.0/cartographer_ros/ros_log_sink.cc:55(ScopedRosLogSink::send)] [topics: /rosout, /tf, /A2/submap_list, /A2/trajectory_node_list, /A2/landmark_poses_list, /A2/constraint_list, /A2/scan_matched_points2, /statistics] W0723 06:29:12.000000 93359 tf_bridge.cc:52] "A2" passed to lookupTransform argument target_frame does not exist. "A2" is the vehicle's name, so it should be able to find itself in the tf tree. This vehicle never started moving because of this issue. |
I have occasionally been seeing robots not starting correctly, and have been unsure about the cause.
I just had this issue reproduce locally with the Marble HD2 robot, and decided to dig in to take a closer look. With two identical robots (one started correctly and one did not), I ran
rosrun tf view_frames
in each solution container to compare the tf graphs. It turns out the robot which did not start correctly is missing thebase_link
tf frame.I'll dig in a bit more if I can, as this is still running locally. I'll post an update here if I find out anything else.
The text was updated successfully, but these errors were encountered: