So far...

Year 1

Posted by Daniela Rato on September 9, 2021

Calibration in simulation mode

the first two weeks of september were focused on the calibration in simulation mode.

this covers:

  • solving previous issues related to the simulation, including: correcting the xacro tf tree, solving problems relative with the robotiq tool, correcting file paths relative to the ur10e simulation files
  • recording a rosbag with the necessary sensor information. this topic was not straightfoward due to the gazebo real time factor being really low, which caused the rosbags to have a much faster time than the real time. the conclusion was that the use_sim_time flag can not be false (false = records in real time), because the static tfs were assuming the real time and the non static tfs were being calculated in real time by atom, which caused different timestamps and throws an error in conclusion, the rosbag needs to be recorded with the use_sim_time=true and the bag_rate used to play the bag needs to be around 0.1 .
  • collecting data in the form of collections, then used to calibrate. here there were also several problems. first the charuco wasn't being detected by the cameras, which was due to the errors in the pattern detector. then atom wasn't generating the interactive markers to create the data lablers for all the lidar sensors and the save collection menu wasn't showing. there was also a mismatch between the image of the charuco in simulation and its lidars projections. all of these issues were addressed and are now solved.
  • the actual calibration process. this process is throwing several errors. still not solved.

Issues

calibration in simulation - open

use_sim_time argument - closed

robotiq crashes in gazebo - closed

mismatch between actual charuco and camera and lidar projections - closed

adjust configure_calibration_pkg to one interactive marker per sensor - closed

menu to save collection - closed

error due to -nig flag - open