# GR-VIP：用于精确视觉运动策略的几何表示条件化

> 原文链接：https://robotics.hellobike.com/papers/2026-10-16-2610-22997
> 频道：papers | 更新日期：2026-10-16 | 标签：arxiv, cs.RO
> 核心摘要：GR-VIP利用相对相机位姿预训练视觉编码器的稠密关系特征，解决移动机器人精确目标相对控制问题，并在真实对接任务中验证。

# GR-VIP：用于精确视觉运动策略的几何表示条件化

::: v-pre
> 原文链接：[GR-VIP: Geometric representation conditioning for precise visuomotor policy](https://arxiv.org/abs/2610.22997)（arXiv 2610.22997 · 发布日期：2026-10-16）

**GR-VIP: Geometric representation conditioning for precise visuomotor policy**

GR-VIP利用相对相机位姿预训练视觉编码器的稠密关系特征，解决移动机器人精确目标相对控制问题，并在真实对接任务中验证。

## 核心要点

- 精确终端位置与朝向对齐对学习式移动机器人控制仍具挑战。
- GR-VIP不直接使用最终相对位姿，而以位姿回归前的稠密跨视图关系特征条件化动作预测。
- 在真实对接任务中，训练环境和未见环境均实现高对接成功率与毫米级终端精度，迁移无需额外训练。
- 补充实验显示，该策略可对齐非对接目标，并在测试的低光照条件下完成对接。

**验证范围与局限**：验证范围为真实世界对接任务，另含非对接目标对齐和一种低光照条件的补充实验。

## 英文摘要 (Abstract)

Achieving precise terminal position and heading alignment remains a significant challenge for learning-based mobile robot control. Recent vision-based learning methods have shown promise for precise target-relative control and generalization to unseen environments with pretrained visual encoders, but general-purpose visual representations are not specifically designed to capture the fine-grained spatial differences between current and goal views required for precise control, and real-world demonstrations combining millimeter-scale terminal accuracy with zero-shot transfer across environments remain limited. To address these challenges, we propose Geometric Representation Conditioning for Precise Visuomotor Policy (GR-VIP), a goal-conditioned visuomotor policy that leverages a visual encoder pretrained for relative camera pose estimation. Rather than using the final relative-pose estimate, GR-VIP conditions action prediction on the dense relational features produced before pose regression, preserving fine-grained cross-view information for learning corrective actions from demonstrations. We evaluate GR-VIP on a real-world docking task and demonstrate high docking success rates and millimeter-scale terminal accuracy in both the training and unseen environments, without additional training for transfer. Furthermore, supplementary experiments illustrate the policy's ability to align with non-docking targets and complete docking under a tested low-light condition.

作者：Chanhwi Lee, Jiwon Seong, 최은제, Hohyung Lee, Jongsoo Lee, Junyeong Bae, Inveom Kwak, Woosang Shin, Soohee Han
:::

## 原始参考来源 (Sources)

- [arXiv 2610.22997：GR-VIP: Geometric representation conditioning for precise visuomotor policy](https://arxiv.org/abs/2610.22997)