How to Get Human Demonstration Data in Mujoco about act

[tonyzhaozh]

Yes that part of the code is not released. What we did is to use the physical ALOHA leader arms to teleoperate the simulated ALOHA follower arms. The operator watches the rendering of the mujoco env on a monitor.

Check out this function to get a rough sense of it. Note it will not run without ALOHA hardware. https://github.com/tonyzhaozh/act/blob/main/sim_env.py#L246

[Jessy-Huang]

Yes that part of the code is not released. What we did is to use the physical ALOHA leader arms to teleoperate the simulated ALOHA follower arms. The operator watches the rendering of the mujoco env on a monitor.

Check out this function to get a rough sense of it. Note it will not run without ALOHA hardware. https://github.com/tonyzhaozh/act/blob/main/sim_env.py#L246

What is the format of source of data(https://github.com/tonyzhaozh/act/blob/main/sim_env.py#L253，https://github.com/tonyzhaozh/act/blob/main/sim_env.py#L255 )，We have a robotic arm in our lab that also has cameras, and based on your experimental process, we thought we could link the Mujoco emulator to our robotic arm and use the data from our lab's robotic arm to drive the Mujoco emulator to acquire the corresponding data.

Can you post the format of master_bot_left and master_bot_right that you ran out and I'll get the corresponding data on my robotic arm