Esacode

Docs / Demo projects

Booster landing

A rocket booster learns to land on a drone ship by reinforcement learning on your GPU, then launches and lands live in 3D with onboard cameras.

A Falcon 9-class booster learns to land on a drone ship at sea: thousands of boosters train at once on the GPU, lesson by lesson. Then a full flight runs live at Earth scale: lift-off, separation, the flip, the entry burn and the landing, with cameras on board, sound and a commentator.

The flows

FlowWhat it does
1 · Learn to landThousands of boosters at once on the GPU learn to land, lesson by lesson; one attempt is replayed in 3D every few seconds. The policy is saved to a dataset on your worker; publish a version of it to GitHub when you are happy with it (the project's Data tab).
2 · Launch and landPress Launch: lift-off, separation, flip, entry burn, then the learned policy lands the booster on the drone ship, live in 3D with onboard cameras.

What you need

An NVIDIA GPU for training. Flow 2 runs on any worker with a policy from flow 1. The worker needs Run code from projects turned on (Computing > your worker > Worker settings).

Try it

  1. Open Projects > Demos and choose Booster landing. Read its readme.
  2. Open flow 1 in the Blueprint and press Run: it runs as it is on your own workers. Switch to the View page to watch it. Its runs and results are yours only.
  3. To change it, choose Clone into my workspace: a copy goes into your workspace, with its readme, blueprints, the code of its blocks and its files, saved to a GitHub repository of yours (link GitHub first, under Profile > GitHub).

Learn more

3D scenes and rendering.