# Outcome-Conditioned End-Effector Geometry Across Vision-Language-Action Policies

arXiv: https://arxiv.org/abs/2609.21659 · HTML: https://mlsanity.com/paper/2609.21659/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Failure recovery and robustness, one of 13 directions in Vision-language-action models.
- Builds on 19 of its 23 references that are in the map.
- First posted on arXiv 18 Sep 2026.

## Placed in

- Vision-language-action models / [Failure recovery and robustness](https://mlsanity.com/vla/failure-recovery-and-robustness/)

## Builds on (most cited first, 19 of 19)

- [OpenVLA: An Open-Source Vision-Language-Action Model](https://mlsanity.com/paper/2406.09246/) (2024)
- [π0.5: a Vision-Language-Action Model with Open-World Generalization](https://mlsanity.com/paper/2504.16054/) (2025)
- [LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learning](https://mlsanity.com/paper/2306.03310/) (2023)
- [Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Success](https://mlsanity.com/paper/2502.19645/) (2025)
- [UniVLA: Learning to Act Anywhere with Task-centric Latent Actions](https://mlsanity.com/paper/2505.06111/) (2025)
- [SAFE: Multitask Failure Detection for Vision-Language-Action Models](https://mlsanity.com/paper/2506.09937/) (2025)
- [VLATest: Testing and Evaluating Vision-Language-Action Models for Robotic Manipulation](https://mlsanity.com/paper/2409.12894/) (2024)
- [RoboEval: Where Robotic Manipulation Meets Structured and Scalable Evaluation](https://mlsanity.com/paper/2507.00435/) (2025)
- [When Vision Overrides Language: Evaluating and Mitigating Counterfactual Failures in VLAs](https://mlsanity.com/paper/2602.17659/) (2026)
- [Evaluating Uncertainty and Quality of Vision-Language-Action-enabled Robots](https://mlsanity.com/paper/2507.17049/) (2025)
- [LIBERO-Para: A Diagnostic Benchmark and Metrics for Paraphrase Robustness in VLA Models](https://mlsanity.com/paper/2603.28301/) (2026)
- [Hide-and-Seek in Trajectories: Discovering Failure Signals for VLA Runtime Monitoring](https://mlsanity.com/paper/2605.30834/) (2026)
- [What Are We Actually Benchmarking in Robot Manipulation?](https://mlsanity.com/paper/2606.04233/) (2026)
- [NEBULA: Do We Evaluate Vision-Language-Action Agents Correctly?](https://mlsanity.com/paper/2510.16263/) (2025)
- [Experiences from Benchmarking Vision-Language-Action Models for Robotic Manipulation](https://mlsanity.com/paper/2511.11298/) (2025)
- [Distracted Robot: How Visual Clutter Undermine Robotic Manipulation](https://mlsanity.com/paper/2511.22780/) (2025)
- [When Does Language Matter? Multilingual Instructions Reveal Step-wise Language Sensitivity in Vision-Language-Action Models](https://mlsanity.com/paper/2606.11906/) (2026)
- [Tri-Info: Generalizable, Interpretable Failure Prediction for VLA Models via Information Theory](https://mlsanity.com/paper/2606.19998/) (2026)
- [Capability and Robustness Cannot Both Be Free: An Information-Theoretic Bound for Vision-Language-Action Models](https://mlsanity.com/paper/2605.25889/) (2026)

## Abstract (arXiv)

Vision-language-action (VLA) policies solve the same manipulation task through different action interfaces, but task success alone does not establish whether their physical executions agree. We study cross-policy end-effector geometry in 15,000 closed-loop LIBERO rollouts from four policies. The primary clean-condition analysis forms 3,600 configuration-matched, and therefore dependent, policy pairs. Both-success pairs have a median normalized dynamic time warping distance of 0.0120 m versus 0.0380 m when exactly one policy succeeds. This ordering holds in every task, every policy pair, and nine sampling and band-limited representations; however, the ratio varies severalfold across representations, so we report the direction rather than a fixed multiple. Both-failure pairs are more separated again but rest on thin, uneven support, so we report them as exploratory. Within successful executions, partner replacements separate more across tasks than across initial states. A matched baseline still reveals measurable, heterogeneous residual policy differences, so a low cross-policy distance does not imply interchangeability. Successful executions sit about as far from same-task demonstrations as those demonstrations sit from each other, compatible with task-associated geometry without separating training-data overlap from task constraints. A common 72-action window preserves the ordering but reduces its magnitude; endpoint and duration adjustment likewise leaves a positive mixed-outcome coefficient relative to both-success pairs, though its magnitude is specification-dependent. Under composite visual stress, policy rankings and pair composition change together.