RoboticsGPU8 credits

MuJoCo — Rigid-Body Robot Simulation

Rollout policies against MuJoCo and PyBullet environments

MuJoCo went open-source in 2022 and is now the most-used rigid-body physics engine in robotics research. SciRouter exposes MuJoCo (and PyBullet as a fallback) as a single API call: pass an environment name and a policy, get back joint-position trajectories, rewards, and episode summaries. Skip the dependency hell of CUDA drivers, GLFW, and OpenGL contexts.

$0.08
per API call
8
credits per call
/v1/robotics/simulate
API endpoint

Features

MuJoCo + PyBullet environments
Returns per-step joint positions, rewards, success flag
Trajectory preview for visualization
Deterministic seeds for reproducibility

Quick Start

MuJoCo — Python Examplepython
import requests

r = requests.post(
    "https://scirouter-gateway-production.up.railway.app/v1/robotics/simulate",
    headers={"Authorization": "Bearer sk-sci-..."},
    json={
        "environment": "cartpole-v1",
        "policy": "random",
        "episodes": 5,
        "max_steps": 500,
    },
)
result = r.json()["data"]
print(f"Mean reward: {result['mean_reward']:.2f}")
print(f"Success rate: {result['success_rate']:.0%}")

Use Cases

1

Weekend robotics research

2

University lab simulation without GPU setup

3

Benchmarking RL policies

4

Teaching embodied AI

Learn More

MuJoCo and LeRobot Simulation API for Robotics Research

13 min read

Related Tools

RFdiffusion

Generate novel protein backbones and enzyme scaffolds via denoising diffusion

$0.12/call

Start Using MuJoCo

500 free credits every month. No credit card required.

Back to all tools