Joint pos observation not biased by encoder domain randomization in default environments #1065
Closed
jgueldenstein
started this conversation in
General
Replies: 2 comments
-
|
It's a bug, we had a fix in this PR but I never merged it. |
Beta Was this translation helpful? Give feedback.
0 replies
-
|
Fixed at head. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Hey everyone,
I was wondering if there is a specific reasoning why the joint_rel observation is not biased tracking env.
mjlab/src/mjlab/tasks/velocity/velocity_env_cfg.py
Lines 91 to 94 in efdcadc
The domain randomization on the encoder bias:
mjlab/src/mjlab/tasks/velocity/velocity_env_cfg.py
Lines 248 to 255 in efdcadc
Is this intended?
Beta Was this translation helpful? Give feedback.
All reactions