剪枝后再规划：用于具身问答中稳定前沿探索的分步校准
Prune-Then-Plan: Step-Level Calibration for Stable Frontier Exploration in Embodied Question Answering
AI总结：
Prune-Then-Plan通过分步校准稳定具身问答中的前沿探索，提升导航效率和回答质量。
摘要：
大型视觉-语言模型（VLMs）通过提供强大的语义先验来改进具身问答（EQA）智能体，从而增强了开放词汇推理能力。然而，当直接用于分步探索时，VLMs常常表现出前沿震荡现象，这种不稳定的来回运动是由过度自信和校准错误引起的，导致导航效率低下和回答质量下降。我们提出了Prune-Then-Plan，一种简单而有效的框架，通过分步校准来稳定探索。与信任原始VLM分数不同，我们的方法使用受Holm-Bonferroni启发的剪枝程序剪枝不合理的前沿选择，然后将最终决策委托给基于覆盖度的规划器。这种分离通过依赖人类水平的判断将过度自信的预测转化为保守、可解释的动作，从而校准VLMs的分步行为。在3D-Mem EQA框架中集成后，我们的方法在视觉基础SPL和LLM-Match指标上分别比基线提高了最多49%和33%。总体而言，我们的方法在OpenEQA和EXPRESS-Bench数据集上，在相等的探索预算下实现了更好的场景覆盖。
英文摘要：
Large vision-language models (VLMs) have improved embodied question answering (EQA) agents by providing strong semantic priors for open-vocabulary reasoning. However, when used directly for step-level exploration, VLMs often exhibit frontier oscillations, unstable back-and-forth movements caused by overconfidence and miscalibration, leading to inefficient navigation and degraded answer quality. We propose Prune-Then-Plan, a simple and effective framework that stabilizes exploration through step-level calibration. Instead of trusting raw VLM scores, our method prunes implausible frontier choices using a Holm-Bonferroni inspired pruning procedure and then delegates final decisions to a coverage-based planner. This separation converts overconfident predictions into conservative, interpretable actions by relying on human-level judgments to calibrate the step-level behavior of VLMs. Integrated into the 3D-Mem EQA framework, our approach achieves relative improvements of up to 49% and 33% in visually grounded SPL and LLM-Match metrics respectively over baselines. Overall, our method achieves better scene coverage under equal exploration budgets on both OpenEQA and EXPRESS-Bench datasets.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
