Is it normal to have large linear velocity tracking errors in MuJoCo simulations?

[cyugai]

Hi, thank you for your great work.

I was running play_amp.py and set command[0] = 1. However, I found that the actual velocity in the x direction is around 0.5, which is much lower than the expected value of 1. I obtained the actual x velocity through data.qvel[0].

Additionally, when the robot walks along the x-axis, the deviation in the y direction is also quite large—about 0.1 meters for every meter the robot walks along the x-axis.

I’m wondering if such a large tracking error is expected in this case. Could it be due to the sim2sim differences between IsaacGym and MuJoCo?