Hello,

I'm a student and I'm working with ROS and gazebo for simulation. I'm using this package (https://github.com/Sahloul/ar_sys) that uses Aruco Markers and gives poses of the Arucos.

I started with Gazebo and I was getting really high error in the detected Arucos Marker poses, now that I'm working with actual robots and with Kinects and I get almost no error in the measurements of the Aruco poses. They are almost perfect, like 1 degrees of error, comparing to the almost ~10 degree error in the Gazebo simulator.

In Gazebo I used a simulated camera 1980x1082 and in real life I only have a Kinect that is 640x480.

I also added a little bit of noise to the simulated camera in Gazebo.

Who can I justify this differences in the results? I was expecting real life situation to be worst then in Gazebo.

Thx so much for the help! Gazebo is amazing.