Replicator

Isaac Sim Replicator is a collection of extensions, python APIs, workflows, and tools such as Replicator Composer that enable a variety of synthetic data generation tasks.

Interacting with GUI

These extensions provide UI interfaces to common tasks and examples you can interact with

• Editing Semantic Information: Use the Semantics Schema Editor extension to label objects of interest in your scene.

• Visualizing Semantics: Use the Synthetic Data Visualizer extension to quickly view and debug synthetic data for your scene.

• Recording Synthetic Data: Use the Synthetic Data Recorder extension to generate synthetic data via the UI.

Python Workflows

Once you have configured, tested, and validated your scene’s semantic information, the following tutorials provide useful examples for different synthetic generation workflows via Standalone Application or Script Editor python scripts.

• Offline Dataset Generation

• Online Generation

• Offline Pose Estimation Synthetic Data Generation

• Randomization in Simulation – AMR Navigation

• Randomization in Simulation – UR10 Palletizing

Replicator Composer

Replicator Composer is a tool for creating parameterizable offline datasets in Isaac Sim.

• Using Replicator Composer: See the Replicator Composer (Deprecated) tutorial for a walkthrough on how to generate a dataset with Replicator Composer.

• Parameter List: See the Replicator Composer Parameter List manual for a list of dataset generation parameters.

• Detailed Manual: See the Replicator Composer Manual manual for writing an input file with the dataset generation parameters.

Scene Generation with SceneBlox

• Using SceneBlox for procedural generation: See the Scene Generation with SceneBlox to learn how to use wavefunction collapse with SceneBlox to generate procedural scenes.

• SceneBlox manual: see the Replicator SceneBlox Manual for a more detailed examination.

Synthetic Data Augmentation

Augmenting Synthetic Data: See the Data Augmentation examples on how to augment annotator data using numpy (CPU) or warp (GPU).

Custom Randomization Snippets

Custom Randomization Snippets: See the Randomization Snippets examples on how to use custom randomization snippets to generate synthetic data.