Skip to content

Stanford-TML/TML-Humanoid-Deploy

Repository files navigation

TML Humanoid Deploy

Example experiment folder:

- experiment_folder_name
    - experiment.yaml (copy and extend from existing experiments)
    - policy.onnx (obtained from logs/rsl_rl/... in training code base)
    - ref_motion.npz (obtained from artifacts in training code base)

Run sim2sim with privilege info

python run_controller.py --use_sim --config exported_policies/sirui_test/experiment.yaml

Run sim2real(sim) with foot odometer as velocity and position estimator

# Terminal 1
python run_controller.py (--use_sim) --config exported_policies/sirui_test/experiment.yaml --use_odom
# Terminal 2
python run_state_estimation.py --use_sim --visualize

Note sim2real requires running lidar slam code link

Run sim2real(sim) with SLAM fusing with foot odom

python run_controller.py (--use_sim) --config exported_policies/sirui_test/experiment.yaml --use_odom --use_slam
python run_state_estimation.py --use_sim --visualize --use_slam

Some note

  1. Sim2Real code remain untested
  2. Anchor link is reset to pelvis instead of default torso_link.
  3. If you have any question please contact Sirui Chen ericcsr@stanford.edu

About

WIP deployment code for TML codebase

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors