ONNX Model Zoo

diasAiMaster/unitree-go2-velocity-flat

By diasAiMaster · ↓ 0 · ♥ 1 · Updated 2026-02-07

What's in this repo

This page lists the files in diasAiMaster/unitree-go2-velocity-flat, including policy.onnx.

  • 1 ONNX models (.onnx files)
  • 26 files, including 11 .log, 6 .pt, 2 .0, 2 .yaml, 1 .md, 1 .diff
  • Last updated 2026-02-07

ONNX files

File Size Download
policy.onnx — Download

How to use

These ONNX files can be loaded with ONNX Runtime. To inspect a model graph, upload the file to the Netron viewer. Download links point to the original files on Hugging Face.

README

Unitree Go2 — Velocity Flat (PPO)

RL locomotion policy for the Unitree Go2 quadruped robot, trained on flat terrain using PPO.

Demo

Unitree Go2 RL Locomotion in MuJoCo

Training

  • Framework: unitree_rl_mjlab (MuJoCo Warp)
  • Task: Mjlab-Velocity-Flat-Unitree-Go2
  • Algorithm: PPO (RSL-RL)
  • Hardware: 10× NVIDIA RTX A4000, 56 CPU cores
  • Environments: 8192 parallel
  • Training time: ~18 minutes (506 iterations)

Results

Metric Value
Mean reward 52.9
Mean episode length 1000 (max, no falls)
Steps/sec 628K-738K

Files

File Description
policy.onnx + policy.onnx.data ONNX model for deployment (go2_ctrl)
model_500.pt Final PyTorch checkpoint (best for fine-tuning)
model_0.pt ... model_400.pt Intermediate checkpoints every 100 steps
params/env.yaml Environment configuration
params/agent.yaml Agent/PPO configuration
events.out.tfevents.* TensorBoard training logs

Usage

Deploy in MuJoCo simulator

# Copy ONNX model
cp policy.onnx policy.onnx.data \
  unitree_rl_mjlab/deploy/robots/go2/config/policy/velocity/v0/exported/

# Run simulator + controller
cd unitree_mujoco/simulate/build && ./unitree_mujoco
cd unitree_rl_mjlab/deploy/robots/go2/build && ./go2_ctrl --network=lo

Fine-tune on rough terrain

# Place model_500.pt in logs/rsl_rl/go2_velocity/<run_name>/
python scripts/train.py Mjlab-Velocity-Rough-Unitree-Go2 \
  --agent.resume=True \
  --agent.load-run="<run_name>" \
  --agent.load-checkpoint="model_500.pt" \
  --agent.algorithm.learning-rate=1e-4

Known Issues

The upstream unitree_rl_mjlab has bugs that crash multi-GPU training on rough terrain — see Issue #9 and PR #8.

Model info

reinforcement-learning rsl-rl
rsl-rl tensorboard onnx reinforcement-learning robotics locomotion unitree go2 mujoco ppo license:bsd-3-clause region:us
↓ 0 ♥ 1

Created: 2026-02-06

Updated: 2026-02-07

View on Hugging Face

Files (26)

.gitattributes
README.md
events.out.tfevents.1770373110.brev-r7jhvq2ew.27241.0
git/unitree_rl_mjlab.diff
model_0.pt
model_100.pt
model_200.pt
model_300.pt
model_400.pt
model_500.pt
params/agent.yaml
params/env.yaml
policy.onnx ONNX
policy.onnx.data
runs/flat/events.out.tfevents.1770373110.brev-r7jhvq2ew.27241.0
torchrunx/2026-02-06T10:17:31/localhost.log
torchrunx/2026-02-06T10:17:31/localhost[0].log
torchrunx/2026-02-06T10:17:31/localhost[1].log
torchrunx/2026-02-06T10:17:31/localhost[2].log
torchrunx/2026-02-06T10:17:31/localhost[3].log
torchrunx/2026-02-06T10:17:31/localhost[4].log
torchrunx/2026-02-06T10:17:31/localhost[5].log
torchrunx/2026-02-06T10:17:31/localhost[6].log
torchrunx/2026-02-06T10:17:31/localhost[7].log
torchrunx/2026-02-06T10:17:31/localhost[8].log
torchrunx/2026-02-06T10:17:31/localhost[9].log

Related models

More from diasAiMaster

No other models by this author.

More reinforcement-learning models

Browse all reinforcement-learning models

Popular models