Rocket Forge

Robotics AI Platform Documentation

Rocket Forge is the operating system for robotics AI development. It centralizes simulation, synthetic data generation, VR teleoperation, and foundation model training into one unified platform.

Quick Start GitHub

Core Concepts

Simulation

Orchestrate Isaac Sim environments for robot training and testing.

Synthetic Data

Generate diverse training datasets with domain randomization.

VR Teleoperation

Capture expert demonstrations through immersive VR control.

Training

Train and fine-tune foundation models with managed infrastructure.

Installation

Install via pip

$ pip install rocket-forge

Clone from source

$ git clone https://github.com/satori-dynamics/rocket-forge.git
$ cd rocket-forge
$ pip install -e .

Basic Usage

Create a simulation environment

from rocket_forge import Simulation, Robot, Environment

# Create environment
env = Environment(
    name="tabletop-manipulation",
    physics_dt=1/120,
    rendering_dt=1/30
)

# Add robot
robot = Robot(
    urdf_path="robots/franka/franka.urdf",
    position=[0, 0, 0]
)
env.add(robot)

# Run simulation
sim = Simulation(env)
sim.start()

# Execute actions
for step in range(1000):
    action = policy(sim.get_observation())
    sim.step(action)

Simulation Guide

Continue