Skip to content

Latest commit

 

History

History
197 lines (134 loc) · 5.45 KB

sim-to-sim-locomotion.md

File metadata and controls

197 lines (134 loc) · 5.45 KB

Sim-to-Sim Locomotion Workflow

See also: Inference & Deployment Guide for all deployment options

This guide provides a complete workflow for running locomotion policies in MuJoCo simulation.

Overview

The sim-to-sim workflow allows you to replay IsaacSim/IsaacGym-trained locomotion checkpoints inside MuJoCo for evaluation and testing.

Prerequisites

  • MuJoCo environment set up (scripts/source_mujoco_setup.sh)
  • Holosoma inference environment set up (scripts/source_inference_setup.sh)
  • ONNX model checkpoint
  • Keyboard or joystick for control

Note: Always use --task.interface lo (loopback) when inference and MuJoCo run on the same machine.

Robot Workflows

Unitree G1 (29-DOF)

1. Start MuJoCo Environment

In one terminal, launch the MuJoCo environment:

For keyboard control:

source scripts/source_mujoco_setup.sh
python src/holosoma/holosoma/run_sim.py robot:g1-29dof

For joystick control:

source scripts/source_mujoco_setup.sh
python src/holosoma/holosoma/run_sim.py robot:g1-29dof \
    --simulator.config.bridge.enabled=True \
    --simulator.config.bridge.use-joystick=True

The robot will spawn in the simulator, hanging from a gantry.

2. Launch the Policy

In another terminal, run the policy inference:

For keyboard control:

source scripts/source_inference_setup.sh
python3 src/holosoma_inference/holosoma_inference/run_policy.py inference:g1-29dof-loco \
    --task.model-path src/holosoma_inference/holosoma_inference/models/loco/g1_29dof/fastsac_g1_29dof.onnx \
    --task.no-use-joystick \
    --task.interface lo

For joystick control:

source scripts/source_inference_setup.sh
python3 src/holosoma_inference/holosoma_inference/run_policy.py inference:g1-29dof-loco \
    --task.model-path src/holosoma_inference/holosoma_inference/models/loco/g1_29dof/fastsac_g1_29dof.onnx \
    --task.use-joystick \
    --task.interface lo

3. Deploy the Robot

  • In MuJoCo window, press 8 to lower the gantry until robot touches ground
  • In MuJoCo window, press 9 to remove gantry

4. Start the Policy

In policy terminal, press ] (or A button on joystick) to activate the policy.

5. Enter Walking Mode

In policy terminal, press = (or Start button on joystick) to enter walking mode.

6. Control the Robot

  • Use keyboard (w a s d for linear, q e for angular) or joystick to control velocity
  • Left joystick: Move forward/backward/left/right
  • Right joystick: Turn left/right

Booster T1 (29-DOF)

1. Start MuJoCo Environment

In one terminal, launch the MuJoCo environment:

For keyboard control:

source scripts/source_mujoco_setup.sh
python src/holosoma/holosoma/run_sim.py robot:t1-29dof-waist-wrist

For joystick control:

source scripts/source_mujoco_setup.sh
python src/holosoma/holosoma/run_sim.py robot:t1-29dof-waist-wrist \
    --simulator.config.bridge.enabled=True \
    --simulator.config.bridge.use-joystick=True

The robot will spawn in the simulator, hanging from a gantry.

2. Launch the Policy

In another terminal, run the policy inference:

For keyboard control:

source scripts/source_inference_setup.sh
python3 src/holosoma_inference/holosoma_inference/run_policy.py inference:t1-29dof-loco \
    --task.model-path src/holosoma_inference/holosoma_inference/models/loco/t1_29dof/ppo_t1_29dof.onnx \
    --task.no-use-joystick \
    --task.interface lo

For joystick control:

source scripts/source_inference_setup.sh
python3 src/holosoma_inference/holosoma_inference/run_policy.py inference:t1-29dof-loco \
    --task.model-path src/holosoma_inference/holosoma_inference/models/loco/t1_29dof/ppo_t1_29dof.onnx \
    --task.use-joystick \
    --task.interface lo

3. Deploy the Robot

  • In MuJoCo window, press 8 to lower the gantry until robot touches ground
  • In MuJoCo window, press 9 to remove gantry

4. Start the Policy

In policy terminal, press ] to activate the policy.

5. Enter Walking Mode

In policy terminal, press = to enter walking mode.

6. Control the Robot

Use keyboard to control velocity:

  • w a s d for linear movement (forward/backward/left/right)
  • q e for angular movement (turn left/right)

MuJoCo Controls Reference

Enter these commands in the MuJoCo window (not the policy terminal):

Gantry Controls

  • 7: Lift the gantry
  • 8: Lower the gantry
  • 9: Disable/remove the gantry

General Controls

  • Backspace: Reset simulation

Policy Controls Reference

Enter these commands in the policy terminal (where you ran run_policy.py):

General Controls

Action Keyboard Joystick
Start the policy ] A button
Stop the policy o B button
Set robot to default pose i Y button

Locomotion Controls

Action Keyboard Joystick
Switch walking/standing = Start button
Adjust linear velocity w a s d Left stick
Adjust angular velocity q e Right stick

Tips and Troubleshooting

  • Reset anytime: Press Backspace in the MuJoCo window to reset the simulation
  • Interface: Always use lo (loopback) for sim-to-sim on the same machine
  • Standing mode: The robot starts in standing mode - press = or Start to switch to walking mode
  • Default pose: Press i or Y button to return the robot to standing pose