Week #14 11/05 - 15/05

Rosbag merge

Posted by Rui Costa on May 14, 2020

This weeks' main objective was to continue writting the disseration. Aside from that, in order to playback the data from the ATLASCAR2 steering wheel angle and velocity telemetry as well as the LIDAR sensor data at the same time I merged the two bagfile using a python code available on github. Despite the rosbag info displaying no errors for the merged bagfile I couldn't get it to publish the tf frames of the car. Since this tf tree is essential for both the GNN and VO nodes to work I tried to understand why this error was appearing, whith no luck unfortunatly. So, in order to go around this problem I played back the data from a working bagfile wich publishes this data and then paused it after the tf message was published, so that both nodes could access it. Then by simply starting the merged bagfile, both nodes were able to work.

Despite me thinking that the visual lag between the ATLASCAR short term path and the LIDAR sensor data was due to the fact that the data was being published by two seperate bagfiles, the same happened with the merged bagfile. This now leads me to believe that this lag is due to computer performance or even due to the VO node's extreme ammount of displayed information bottlenecking Rviz.