More accurate time stamps for poses

[NikolausDemmel]

Is it possible from ARNL to access a more accurate time stamp for the laser pose. At the moment the current time in the 10Hz loop callback in used (

ros-arnl/rosarnl_node.cpp

Line 176 in 19702d6

pose.header.stamp = ros::Time::now();

), however this is likely not very accurate. E.g. after a laser update of the localization filter, the timestamp of the current pose should be that time when the laser data used in the update was taken by the sensor.

There is a similar issue for the sensor data itself. If we want to add e.g. publishing the laser data to ROS, we currently would also need to use the current time in the loop callback (as done here: https://github.com/TRECVT/rosaria/blob/9e00c4e78bc9e2bb03cdeaf69c194898c1715ba9/src/RosAria.cpp#L678). Is there a way to get the acquisition time of laser scans?

[reed-adept]

I just updated rosarnl_node to use ARNL's timestamp for when it last re-localized (see commit 4c72a62). This is not exactly the time when the sensor data was acquired, since ARNL is asynchronously getting sensor data from various sources, but it should represent the delay between ARNL determining a new localized pose and when rosarnl_node publishes that pose as amcl_pose. If there's a specific need for something more than that, I don't know if it can be done with the existing ARNL API but let me know so we can look into it.