Getting Started

Simulo is a cloud SDK for simulation-first robot learning: you write a task in Python against a small, stable API, and simulo run submits it to the Simulo cloud, where it runs on GPU workers. Use the CLI or Simulo Console to follow the job and retrieve its outputs.

The starter app already pins a validated cartpole from the Simulo global catalog. That keeps the first path short: install, create, log in, run, and observe. Explore or publish assets after that first success, and when the policy is good, simulo export converts it to an ONNX bundle you can run off the platform.

If you want the fastest path from zero to a trained policy, start with the Quickstart. Otherwise, work through the pages below in order.

QuickstartInstall, scaffold an app, log in, run it in the cloud, then observe or stop the job.

Installpip install simulo. Python 3.11+, torch-free, with simulator-free USD/URDF packaging tools included.

Create your first appsimulo create scaffolds a runnable app folder from a starter template — no network, no auth.

Log insimulo login signs the CLI into the Simulo cloud — browser sign-in, with a headless fallback.

Run & submitsimulo run app.py resolves and pins catalog assets, submits the app to the cloud, and follows its logs.

ObserveUse the CLI or Console to follow logs, stop a job, inspect results, and download models, recordings, or outputs.

Choose & validate assetsAfter your first run, browse validated robots or validate and publish your own USD/URDF package.

Continue trainingsimulo run --from picks up training from a finished job's checkpoint instead of starting over.

Export a policy to ONNXsimulo export converts the trained policy to a portable ONNX bundle and ships a verifier that proves it runs with Simulo uninstalled.