SAGE: Scalable Agentic 3D Scene Generation for Embodied AI

TL;DR

SAGE is an innovative framework that automatically generates realistic, diverse, and physically valid 3D environments for embodied AI, enabling scalable policy training and generalization to unseen scenarios.

Contribution

It introduces a novel agentic system that understands user tasks and iteratively generates simulation-ready 3D scenes, improving scalability and realism over prior rule-based methods.

Findings

Generated environments are realistic and diverse.

Policies trained on SAGE scenes generalize well to new objects.

SAGE enables scalable simulation data creation for embodied AI.

Abstract

Real-world data collection for embodied agents remains costly and unsafe, calling for scalable, realistic, and simulator-ready 3D environments. However, existing scene-generation systems often rely on rule-based or task-specific pipelines, yielding artifacts and physically invalid scenes. We present SAGE, an agentic framework that, given a user-specified embodied task (e.g., "pick up a bowl and place it on the table"), understands the intent and automatically generates simulation-ready environments at scale. The agent couples multiple generators for layout and object composition with critics that evaluate semantic plausibility, visual realism, and physical stability. Through iterative reasoning and adaptive tool selection, it self-refines the scenes until meeting user intent and physical validity. The resulting environments are realistic, diverse, and directly deployable in modern simulators for policy training. Policies trained purely on this data exhibit clear scaling trends and generalize to unseen objects and layouts, demonstrating the promise of simulation-driven scaling for embodied AI. Code, demos, and the SAGE-10k dataset can be found on the project page here: https://research.nvidia.com/labs/dir/sage/.