# AI use cases for Lyapunov (ManfredCh/Lyapunov)

Repo: [ManfredCh/Lyapunov](https://github.com/ManfredCh/Lyapunov) | Use cases: 4 | Page: https://dedupedai.com/use-cases/manfredch-lyapunov/

Lyapunov is an LLM-first, 3D-native robotics and agent desktop framework that brings scene building, robot preparation, controlled simulation and camera data into one Linux workspace.

This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL.

- Prepare and run a robot action in simulation: https://dedupedai.com/use-cases/manfredch-lyapunov/prepare-and-run-robot-in-simulation/
- Collect RGB-D camera data from a simulated scene: https://dedupedai.com/use-cases/manfredch-lyapunov/collect-camera-data-from-scene/
- Build a collision-ready 3D scene from a GLB file: https://dedupedai.com/use-cases/manfredch-lyapunov/build-collision-ready-scene-from-glb/
- Install Lyapunov on Linux and open a local workspace: https://dedupedai.com/use-cases/manfredch-lyapunov/install-lyapunov-on-linux/

## How to prepare and run a robot action in simulation with Lyapunov

Repo: [ManfredCh/Lyapunov](https://github.com/ManfredCh/Lyapunov) | Difficulty: Intermediate | Time: 90 minutes | Basis: From the README

Set up a robot in a 3D scene, initialize physics, and run one bounded action with feedback.

**Who it is for:** A robotics developer who needs to check a robot's controls and physical response before writing or testing a control policy.

**The problem:** Robot models can look ready in a viewer while their physics world is not, and whole-body behavior is easy to assume incorrectly.

**The outcome:** A robot that is prepared, initialized and ready, with one explicitly executed action and inspected world feedback.

### Steps

1. Install Lyapunov and launch it, then open or create a project in the workbench.
2. Select the robot and the physics engine, then review the available controls, dependencies, placement and collision geometry.
3. Prepare the world while it is paused, then initialize physics and confirm the robot is ready.
4. Run one bounded action with the robot controls, then stop execution and inspect the actual world feedback.

### Prompt to paste into your agent

```text
In the ManfredCh/Lyapunov repo, prepare the robot <robot-name> in a paused world using the default MuJoCo engine, initialize physics, and run one bounded joint action. Report the robot's ready state and the actual world feedback after the action. Verify that the robot was initialized before any action was executed.
```

### Caveat

README says importing a robot such as G1 does not automatically stand it up, and a geometric gait command does not establish stable walking, so expect to verify each behavior yourself.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/manfredch-lyapunov/prepare-and-run-robot-in-simulation/

## How to collect RGB-D camera data from a simulated scene with Lyapunov

Repo: [ManfredCh/Lyapunov](https://github.com/ManfredCh/Lyapunov) | Difficulty: Advanced | Time: 120 minutes | Basis: From the README

Place scene cameras, calibrate, annotate and record data for a dataset where the engine supports it.

**Who it is for:** A research team member who needs labeled camera data from a simulated environment for perception work.

**The problem:** Camera capture, calibration and annotation are spread across tools, and support differs by engine.

**The outcome:** Recorded RGB-D captures with calibration and annotations, exported as a dataset from a supported engine.

### Steps

1. Open the scene in the workbench and add or select scene cameras in the camera panel.
2. Check the engine-specific support for named engine cameras and RGB-D capture in docs/CAPABILITY_MATRIX.md.
3. Run calibration and annotations, then start recording in the workbench.
4. Export the dataset only from an engine that supports it, following docs/SCENE_CAMERA_PIPELINE.md.

### Prompt to paste into your agent

```text
Using the ManfredCh/Lyapunov repo, read docs/CAPABILITY_MATRIX.md and docs/SCENE_CAMERA_PIPELINE.md, then set up RGB-D capture for the scene <scene-name> on the engine <engine-name>. Report which camera types and dataset export options are supported for that engine. Verify support against the capability matrix before recording.
```

### Caveat

README states that the workspace supports these camera tools only where the selected engine supports them, so check the capability matrix for each engine before relying on a dataset.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/manfredch-lyapunov/collect-camera-data-from-scene/

## How to build a collision-ready 3D scene from a GLB file with Lyapunov

Repo: [ManfredCh/Lyapunov](https://github.com/ManfredCh/Lyapunov) | Difficulty: Intermediate | Time: 90 minutes | Basis: From the README

Import a GLB environment, align it, add collision geometry and confirm the physical response.

**Who it is for:** A 3D creator or simulation developer who has a GLB environment and needs it to behave physically in a robot simulation.

**The problem:** A visual asset does not give a robot anything to collide with, and imported scale or orientation can be wrong.

**The outcome:** A saved scene with aligned visual assets and matching static collision geometry that reopens with the same physical response.

### Steps

1. Import the GLB file into the Scene from the workbench.
2. Inspect the imported asset's scale, axes and orientation, and correct them if needed.
3. Prepare static environment collision geometry that matches the room or layout, then apply it to the current world.
4. Save the scene, reopen it, and check the actual world response after the edits.

### Prompt to paste into your agent

```text
In the ManfredCh/Lyapunov repo, import the GLB file <path-to-glb> into a new scene, prepare static collision geometry aligned with it, and apply it to the current world. Report the scale and orientation checks and the saved scene name. Verify the physical result by checking the world response after reopening the scene.
```

### Caveat

README says a visual Gaussian scene needs separately aligned physical geometry for contact, and finite surface thickness can affect narrow openings, so a visually correct scene may still behave wrongly.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/manfredch-lyapunov/build-collision-ready-scene-from-glb/

## How to install Lyapunov on Linux and open a local workspace with Lyapunov

Repo: [ManfredCh/Lyapunov](https://github.com/ManfredCh/Lyapunov) | Difficulty: Beginner | Time: 30 minutes | Basis: From the README

Run the published installer, confirm the launcher, and open a guest workspace without an account.

**Who it is for:** A developer or researcher on a Linux x64 desktop who wants to try Lyapunov before setting up anything else.

**The problem:** Installing a robotics stack usually means matching several runtimes, and it is unclear which parts are needed for a first run.

**The outcome:** A working Lyapunov install with MuJoCo prepared, a launcher on disk, and a local guest workspace open.

### Steps

1. Open a regular terminal in your graphical desktop session.
2. Run `curl -fsSL https://vorynel.com/lyapunov/install.sh | sh`
3. Wait until the installer reports ready, then start the app with `"$HOME/.local/bin/lyapunov"`
4. On the welcome page, choose a language, then choose Try as guest to open a local workspace.

### Prompt to paste into your agent

```text
Using the ManfredCh/Lyapunov repo, install Lyapunov on this Linux x64 machine with the published installer, then confirm that the launcher exists at $HOME/.local/bin/lyapunov. Report whether the installer reported ready and whether the native physics check passed. Verify the launcher path before starting the app.
```

### Caveat

README says the install needs a graphical desktop session and Linux x64, and that Isaac Sim and Newton are not prepared by the default installation.

Risk, credentials: Sign in uses website authorization in your browser, and any model provider key you add belongs to that provider, so store and rotate those credentials yourself.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/manfredch-lyapunov/install-lyapunov-on-linux/