# When would Vision-Proprioception Policies Fail in Robotic Manipulation?

arXiv: https://arxiv.org/abs/2602.12032 · HTML: https://mlsanity.com/paper/2602.12032/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 12 Feb 2026.

## Built on by (most cited first, 8 of 8)

- [Bridge-WA: Predicting Where and How the World Changes for Robotic Action](https://mlsanity.com/paper/2607.02195/) (2026)
- [When Absolute State Fails: Evaluating Proprioceptive Encodings for Robust Manipulation](https://mlsanity.com/paper/2605.13067/) (2026)
- [GeoProp: Grounding Robot State in Vision for Generalist Manipulation](https://mlsanity.com/paper/2607.07101/) (2026)
- [How Should Vision-Language-Action Models Use Proprioceptive State?](https://mlsanity.com/paper/2608.03052/) (2026)
- [NoContactNoWorries: Estimating Contact through Vision and Proprioception for In-Hand Dexterous Manipulation](https://mlsanity.com/paper/2606.24450/) (2026)
- [A Causality-aware Infer-diagnose-refine Framework for Test-time Modality Adaptation in VLA Models](https://mlsanity.com/paper/2607.25516/) (2026)
- [Policy-Induced Hand Priors in Humanoid Dual-Arm Manipulation: Diagnosing and Mitigating Initial-Pose Dependence](https://mlsanity.com/paper/2608.11769/) (2026)
- [Blackout vs. Freeze: Analyzing Physical Failure Modes of VLAs under Camera Faults](https://mlsanity.com/paper/2609.39145/) (2026)

## Abstract (arXiv)

Proprioceptive information is critical for precise servo control by providing real-time robotic states. Its collaboration with vision is highly expected to enhance performances of the manipulation policy in complex tasks. However, recent studies have reported inconsistent observations on the generalization of vision-proprioception policies. In this work, we investigate this by conducting temporally controlled experiments. We found that during task sub-phases that robot's motion transitions, which require target localization, the vision modality of the vision-proprioception policy plays a limited role. Further analysis reveals that the policy naturally gravitates toward concise proprioceptive signals that offer faster loss reduction when training, thereby dominating the optimization and suppressing the learning of the visual modality during motion-transition phases. To alleviate this, we propose the Gradient Adjustment with Phase-guidance (GAP) algorithm that adaptively modulates the optimization of proprioception, enabling dynamic collaboration within the vision-proprioception policy. Specifically, we leverage proprioception to capture robotic states and estimate the probability of each timestep in the trajectory belonging to motion-transition phases. During policy learning, we apply fine-grained adjustment that reduces the magnitude of proprioception's gradient based on estimated probabilities, leading to robust and generalizable vision-proprioception policies. The comprehensive experiments demonstrate GAP is applicable in both simulated and real-world environments, across one-arm and dual-arm setups, and compatible with both conventional and Vision-Language-Action models. We believe this work can offer valuable insights into the development of vision-proprioception policies in robotic manipulation.