Improbable-AI/walk-these-ways

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

1438 stars · 222 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

A sim-to-real reinforcement learning starter kit for the Unitree Go1 quadruped robot, implementing the Walk These Ways (MoB) locomotion controller. It trains PPO policies in NVIDIA Isaac Gym with domain randomization and deploys them on the real robot via the unitree_legged_sdk.

Use cases

• train RL locomotion policies for the Unitree Go1
• deploy a learned walking controller on a quadruped robot
• simulate legged robot training with Isaac Gym
• run a pretrained multiplicity-of-behavior locomotion policy
• transfer simulated robot policies to the real world
• tune gait, footswing, and posture for a quadruped

When to choose

• you have a Unitree Go1 and want a proven sim-to-real locomotion pipeline
• you need a research baseline for legged RL with domain randomization
• you want a single policy that switches between diverse gaits at runtime

When to avoid

• you use a robot other than the Unitree Go1
• you need a maintained, actively updated framework
• you lack an NVIDIA GPU for Isaac Gym training

Facets

library · maturity maintenance

reinforcement-learning simulation robotics machine-learning llm-training robotics reinforcement-learning simulation machine-learning python sim-to-real quadruped unitree-go1 isaac-gym legged-locomotion ppo domain-randomization robot-deployment linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="Improbable-AI/walk-these-ways")