# ActiveAlign bridges active perception and VLA policies via observation-aligned data adaptation

- Source: arXiv cs.RO
- Published: 2026-10-19T04:00:00.000Z
- RoboHot: https://robohot.news/en/items/zn2ede9e0fajowtkf8soh1iv7
- Original: https://arxiv.org/abs/2610.23378

## Summary

The paper proposes ActiveAlign, a data adaptation framework that addresses the mismatch between the initial observation state produced by active camera motion and the manipulation policy's training distribution (first-frame distribution sensitivity): on LIBERO, directly prepending camera repositioning to pi_0.5 reduces the success rate from 46.3% to 11.5%.