V-CAGE: Context-Aware Generation and Verification for Scalable Long-Horizon Embodied Tasks

TL;DR

V-CAGE is a framework that generates physically plausible, semantically aligned datasets for long-horizon embodied tasks by enforcing geometric consistency, hierarchical instruction decomposition, and semantic verification, improving policy success and generalization.

Contribution

The paper introduces V-CAGE, a novel closed-loop system combining geometric consistency, hierarchical task decomposition, and visual language model verification for scalable, high-fidelity embodied task datasets.

Findings

Datasets generated by V-CAGE have higher physical and semantic fidelity.

V-CAGE significantly improves downstream policy success rates.

Semantic verification reduces silent failures in task execution.

Abstract

Learning long-horizon embodied behaviors from synthetic data remains challenging because generated scenes are often physically implausible, language-driven programs frequently "succeed" without satisfying task semantics, and high-level instructions require grounding into executable action sequences. To address these limitations, we introduce V-CAGE, a closed-loop framework for generating robust, semantically aligned manipulation datasets at scale. First, we propose a context-aware instantiation mechanism that enforces geometric consistency during scene synthesis. By dynamically maintaining a map of prohibited spatial areas as objects are placed, our system prevents interpenetration and ensures reachable, conflict-free configurations in cluttered environments. Second, to bridge the gap between abstract intent and low-level control, we employ a hierarchical instruction decomposition module. This decomposes high-level goals (e.g., "get ready for work") into compositional action primitives, facilitating coherent long-horizon planning. Crucially, we enforce semantic correctness through a VLM-based verification loop. Acting as a visual critic, the VLM performs rigorous rejection sampling after each subtask, filtering out "silent failures" where code executes but fails to achieve the visual goal. Experiments demonstrate that V-CAGE yields datasets with superior physical and semantic fidelity, significantly boosting the success rate and generalization of downstream policies compared to non-verified baselines.