用结构化LLM实现零样本人机协作，提升默契与信任。
Structured LLM Reasoning for Zero-Shot Human--Robot Coordination Under Hidden Goals
- 将决策拆解为心智推理、分层规划等五步流程
- 人机协作减少交互步骤，信任评分更高
- 适合需要高效协同的智能机器人场景
我们提出一种面向具有私有目标视图的合作建造任务的结构化大型语言模型（LLM）架构，用于零样本人机协作。在分布式部分可观马尔可夫决策过程（Dec-POMDP）框架指导下，该架构将决策过程分解为：(i) 动作条件下的心智理论（ToM）推理，(ii) 分层规划，(iii) 对话理解，(iv) 动作验证，以及 (v) 基于反馈的重规划。我们在真人参与者实验中对比了该方法与无ToM推理的消融版本及离线训练的多智能体强化学习策略。结果表明，所提方法在交互步数上更少，且交互后信任评分显著高于两个基线。这表明，系统性地分解团队决策问题，利用LLM作为难以计算的推理与规划的可处理代理，并保留传统物理可行性验证，能够同时提升任务协调效率与人类体验。
原文摘要 · Abstract (English)
We present a structured large-language-model (LLM) architecture for zero-shot human--robot coordination in a cooperative construction task with private goal views. Guided by a Dec-POMDP formulation, the architecture decomposes decision-making into (i) action-conditioned Theory-of-Mind (ToM) inference, (ii) hierarchical planning, (iii) conversation interpretation, (iv) action verification, and (v) feedback-based replanning. We compare the proposed method with an ablation without ToM inference and a multi-agent reinforcement-learning policy trained offline over many goal pairs. In human-participant experiments, the proposed method required fewer interaction steps and yielded higher post-interaction trust ratings than both baselines. These results suggest that systematically decomposing the team decision problem, using LLMs as tractable surrogates for otherwise intractable inference and planning computations, and retaining conventional verification for physical feasibility can improve both task coordination and the human experience.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。