# There are two steps to generate the datasets:
## Step1:
```bash
git clone git@github.com:johnny09/GMR.git
cd GMR
python scripts/bvh_to_robot_npz_dataset.py 
--src_folder <path_to_lafan1_bvh_file_folder> \ 
--tgt_folder <path_to_output_npz_file_folder> \  
--format lafan1 \
--robot pnd_adam_sp \
--target_fps 30
```
The generated file is in <path_to_output_npz_file_folder>.

The info in the generated motion data including:

"fps";

"root_pos";

"root_rot";

"dof_pos";

"dof_vel";

## Step2:
This step mainly do two things:

1.convert "dof_pos" and "dof_vel" in the motion data from step1 to "qpos" and "qvel"

2.interpolate the motion data to the target_fps

3.extend the motion data

```bash
git clone git@github.com:johnny09/loco-mujoco.git
cd loco-mujoco
python convert_custom_format_and_extend_new.py \
    -b \
    -i <path_to_input_motion_data_folder> \
    -o <path_to_output_motion_data_folder> \
    -of 50 \ 
    --env PndAdamSp
```