Deploy on booster T1

[Gloriabhsfer]

Thank you for the amazing work on this project.

I have a couple of questions regarding deploying the policy on the Booster T1 robot:

The trained policy for T1 uses a 29-DoF model, while the actual robot only has 23 DoF. Should I apply an action mask during deployment to handle this mismatch?

It seems that all the policy observation calculations are designed for the G1 robot, but the MuJoCo simulation results for T1 are correct. Given this, if I want to deploy the released policy on the physical robot, can I simply use the onboard deployment script as follows?

python rl_policy/loco_manip/loco_manip.py \
  --config=config/t1/t1_29dof_falcon.yaml \
  --model_path=models/falcon/t1_29dof.onnx

Would appreciate any clarification on these points. Thanks again!

[hang0610]

Hi, do sim2sim first. If things look normal in Mujoco, then it should be fine in sim2real.

[Gloriabhsfer]

Hi, do sim2sim first. If things look normal in Mujoco, then it should be fine in sim2real.

Thanks for your reply. And i got one more question, you have done the on board deploy? Is the code on the booster board or master board?