Use LLM to generate pose commands for quadruped locomotion based on natural language instructions.

1. Run the setup script:

   ./setup_conda_env.sh

   You’ll be prompted to name your environment.

2. Activate it:

   conda activate <your_conda_env>

3. Install learning code:

   pip install -e drail_learning/rsl_rl_ashton

We use VSCode-style launch.json files to organize runnable tasks per extension folder (e.g., launch.core.json).

1. Link a launch configuration (e.g., for ashton):

   ./activate_launch.sh research_ashton

2. You can now view and run scripts using the Run tab in VSCode. Alternatively, inspect .vscode/launch.json directly to see the tasks and their arguments.

python drail_extensions/drail_extensions/research_ashton/scripts/play.py --task ashton-Go2-Pose-Play --device cpu --real-time --checkpoint drail_extensions/drail_extensions/research_ashton/resources/pretrained_policy/go2_pose/h_pose_r18/model_best.pt

Ensure that in the file rsl_rl_pose_ppo_cfg.py you set your own wandb_entity.

python drail_extensions/drail_extensions/research_ashton/scripts/train.py --task ashton-Go2-Pose-Train --headless --device cuda