# WEDGE框架：基于策略弱点诊断的导航环境自适应生成

> 原文链接：https://robotics.hellobike.com/papers/2026-10-14-2610-19420
> 频道：papers | 更新日期：2026-10-14 | 标签：auto, arxiv, hf-daily-papers
> 核心摘要：提出WEDGE框架，利用LLM分析预训练导航策略失败轨迹并自动验证弱点特征，据此定向生成高难度训练环境。在ProcTHOR与HM3D基准上，该弱点导向适配法显著优于随机及硬样本挖掘策略，证实失败模式可作为环境构建的有效指导信号。

# WEDGE框架：基于策略弱点诊断的导航环境自适应生成

::: v-pre
> 本文由 AI 根据公开内容整理，可能有误，请以原文为准：[From Failures to Learnable Worlds: Policy-Conditioned Environment Generation for Navigation Agents](https://arxiv.org/abs/2610.19420)（HF Daily Papers · arXiv 2610.19420）

## 核心内容

针对具身导航训练中数据增量不等于有效学习的问题，研究提出WEDGE（Weakness-guided Environment Generation）框架。该方法改变传统环境生成仅关注多样性或真实感的范式，转而以预训练策略的具体失败模式为导向。实验覆盖三种不同适配机制的导航Agent，结果显示WEDGE在ProcTHOR数据集上的性能持续优于匹配难度的随机采样及硬样本挖掘基线，且Uni-NaVid与Gemini模型经适配后在HM3D数据集上也获得泛化增益。

技术实现上，WEDGE利用大语言模型对比成功与失败Rollout，提取可执行的弱点特征并在诊断集上自动验证，依据成功率差距筛选最强弱点映射至环境控制因子。系统通过调节这些因子构建针对性困难场景，支持迭代重诊断以追踪策略演进中的弱点漂移。研究表明不同策略具有独特弱点画像，需定制化训练环境，验证了将策略失败转化为环境生成参数的工程可行性。

## 关键看点

- **弱点特征提取**：利用LLM对比成功/失败轨迹提出可执行弱点，经诊断集自动验证并按成功率差距排序，精准定位策略短板
- **环境定向生成**：将验证后的弱点映射为环境控制因子，动态调整参数构建特定难度区间场景，实现训练数据与策略需求的对齐
- **跨域泛化验证**：在ProcTHOR上显著优于随机与硬样本基线，Uni-NaVid和Gemini适配后在HM3D亦获增益，证明方法通用性
- **迭代适应机制**：支持多轮重诊断，捕捉策略提升过程中主导弱点的持续或转移，为后续环境构建提供动态更新指导

## 工程与产业观察

WEDGE确立了“失败即数据”的具身训练新范式，解决了通用仿真数据与特定策略需求错配的工程痛点。对于算法工程师，该框架提供了从错误日志到合成数据生成的自动化闭环思路；在Sim2Real场景中，这种弱点导向的数据合成策略有望大幅降低对海量无效数据的依赖，提升导航模型在长尾Corner Case下的鲁棒性与收敛效率。

## 英文摘要 (Abstract)

Simulated environments enable scalable training data generation for embodied agents, but more data does not necessarily provide what a particular agent needs to learn. Existing environment-generation pipelines largely optimize generic properties such as realism, diversity, or task coverage, while different pretrained policies can fail for different environmental reasons. We therefore introduce WEDGE, a framework for weakness-guided navigation learning in which the failure modes of a pretrained agent guide the generation of training environments at each adaptation round. Given a pretrained policy, WEDGE uses an LLM to propose executable weakness features from contrasting successful and failed rollouts, automatically validates them on the diagnostic set, and selects the strongest validated weakness based on its success-rate gap. The validated weaknesses are then mapped to the environmental factors that control them. WEDGE varies these factors to construct environments in the difficulty regimes where the policy struggles, producing targeted training experience for adaptation. Across three navigation agents with different adaptation mechanisms, we find that policies exhibit distinct weakness profiles and benefit from different training environments. Weakness-conditioned adaptation consistently outperforms matched random and hard-example alternatives on ProcTHOR, while the resulting gains for Uni-NaVid and Gemini also extend to HM3D after ProcTHOR adaptation. Iterative re-diagnosis further shows that dominant weaknesses can persist or shift as the policy improves, providing updated guidance for subsequent environment construction. Together, these results show that policy failures are not only diagnostic signals, but actionable guidance for constructing training environments that target a policy's current weaknesses.

Hugging Face Daily Papers 热度：7 upvotes
作者：Yue Zhang, Han Lin, Zun Wang, Yidong Huang, Zaid Khan, Mohit Bansal
:::

## 原始参考来源 (Sources)

- [HF Daily Papers · arXiv 2610.19420：From Failures to Learnable Worlds: Policy-Conditioned Environment Generation for Navigation Agents](https://arxiv.org/abs/2610.19420)