isaac-sim-sdg

You configure synthetic data generation pipelines in Isaac Sim by defining YAML-based config files that orchestrate actor simulation, object randomization, and sensor placement. This automates end-to-end generation of labeled vision datasets for AI training.

The Isaac Sim SDG pipeline is used to generate rich, labeled synthetic vision datasets by orchestrating multi-agent indoor environments. It enables researchers to capture annotated data, captions, and physical event logs suitable for model training and evaluation.

Yes, you need Isaac Sim 5.1 or later to run these YAML-based SDG pipelines. It requires a pip-installed Isaac Sim environment to ensure compatibility with the Action and Event Data Generation platform components like IRA, IRO, and RTX sensors.

Yes, you can generate annotated indoor-scene datasets using Isaac Sim by configuring multi-agent environments with physical incidents. The pipeline produces comprehensive outputs including annotations, captions, and event logs for vision AI pipelines.

RTX sensor planning in synthetic data generation works by defining camera placement and sensor configurations within your YAML pipeline. It allows precise control over data capture perspectives for generating labeled visual datasets in virtual worlds.

YAML config definitions for Isaac Sim SDG require strict alignment with Isaac Sim's extension namespace. Limitations include dependency on a pip-installed Isaac Sim 5.1+ environment and compatibility constraints with the IRA, IRO, IRC, and IRI components.