CorridorVLA: Explicit Spatial Constraints for Generative Action Heads via Sparse Anchors

TL;DR

CorridorVLA introduces explicit spatial anchors as physical cues to improve generative action models in vision-language tasks, leading to significant success rate improvements on challenging benchmarks.

Contribution

It proposes a novel method that uses sparse physical anchors to explicitly guide action generation, enhancing interpretability and performance.

Findings

Achieves 3.4% to 12.4% success rate improvements on LIBERO-Plus benchmark.

GR00T-Corr variant reaches 83.21% success rate.

Provides interpretable physical constraints for generative policies.

Abstract

Vision--Language--Action (VLA) models often use intermediate representations to connect multimodal inputs with continuous control, yet spatial guidance is often injected implicitly through latent features. We propose $CorridorVLA$, which predicts sparse spatial anchors as incremental physical changes (e.g., $\Delta$-positions) and uses them to impose an explicit tolerance region in the training objective for action generation. The anchors define a corridor that guides a flow-matching action head: trajectories whose implied spatial evolution falls outside it receive corrective gradients, while minor deviations from contacts and execution noise are permitted. On the more challenging LIBERO-Plus benchmark, CorridorVLA yields consistent gains across both SmolVLA and GR00T, improving success rate by $3.4\%$--$12.4\%$ over the corresponding baselines; notably, our GR00T-Corr variant reaches a success rate of $83.21\%$. These results indicate that action-aligned physical cues can provide direct and interpretable constraints for generative action policies, complementing spatial guidance encoded in visual or latent forms. Code is available at https://github.com/corridorVLA.