SO101-Nexus
Get Started

Quickstart

Run your first simulated SO-101, then choose a task or train a policy.

Complete Installation, then run this Python script. No hardware is required.

Run your first environment

import gymnasium as gym
import so101_nexus.mujoco  # registers the MuJoCo env ids

env = gym.make("MuJoCoPickLift-v1", render_mode="human")
obs, info = env.reset(seed=0)
env.action_space.seed(0)

for _ in range(1000):
    action = env.action_space.sample()
    obs, reward, terminated, truncated, info = env.step(action)
    if terminated or truncated:
        obs, info = env.reset()

env.close()

reset(seed=0) controls the environment state. action_space.seed(0) controls the random actions from action_space.sample(). Use both calls to reproduce this rollout.

Importing so101_nexus.mujoco is what registers the ids with Gymnasium, so it must run before gym.make(). Use render_mode="human" for a live viewer window, or render_mode="rgb_array" to capture frames programmatically.

Try another task

Replace the environment ID in gym.make() to choose another task. Environments lists all IDs, episode lengths, observations, rewards, and success conditions.

Use a different object

Pass a config to change what is on the table. Here the default cube becomes a YCB gelatin box:

import gymnasium as gym
import so101_nexus.mujoco
from so101_nexus import PickConfig, YCBObject

config = PickConfig(objects=YCBObject(model_id="009_gelatin_box"))
env = gym.make("MuJoCoPickLift-v1", config=config, render_mode="human")
obs, info = env.reset(seed=0)
env.close()

YCB assets download automatically on first use and are cached locally.

Train a policy

Follow Record, clone, reinforce to record demonstrations or start with a published dataset. See Backends for CPU batching, GPU training, and physics differences.

Next steps

On this page