Files
chenlin d5e9004f6c
Checks / Layout, translation and Colab checks (push) Has been cancelled
Checks / Ruff (push) Has been cancelled
Checks / Framework tests (native backend, CPU) (push) Has been cancelled
Checks / Portable Rust controller (push) Has been cancelled
docs: add AI agent workflow guide and clean up residual single-robot references
2026-11-05 11:52:07 +08:00
..

Six-legged locomotion example

This synthetic MuJoCo robot has six radial legs and 18 actuated hinges. The coxa yaw joint sweeps each leg horizontally. The femur and tibia pitch joints lift and extend its foot. The body and links use box, capsule and sphere collision geometry with explicitly specified masses and inertias. The model's front is +X and its total mass is 5.14 kg.

profile.json identifies the six named supports and the locomotion intent. stance.json supplies the nominal joint angles and trial PD parameters used for native stance calibration. robot.json supplies joint action scales and contact selections for the scaffold. The assets use the repository's Apache 2.0 license.

python scripts/onboard.py inspect assets/multipod_example \
  --profile assets/multipod_example/profile.json
python scripts/onboard.py calibrate --model assets/multipod_example/multipod.xml \
  --config assets/multipod_example/stance.json --report out/multipod-calibration.json
python scripts/onboard.py scaffold-tier3 --config assets/multipod_example/robot.json \
  --calibration out/multipod-calibration.json --tasks-root out/multipod-scaffold
python scripts/train.py --task multipod_example --backend native --device cpu \
  --num-envs 4 --max-iterations 1 --headless --no-tensorboard

The checked-in robot specification is the calibrated task used by the training command. The scaffold example writes a separate copy for inspection. W/S command forward and reverse, A/D command left and right, and J/L command yaw.