ChatPaper.aiChatPaper

自适应领域建模与语言模型:面向任务规划的多智能体方法

Adaptive Domain Modeling with Language Models: A Multi-Agent Approach to Task Planning

June 24, 2025
作者: Harisankar Babu, Philipp Schillinger, Tamim Asfour
cs.AI

摘要

我们推出TAPAS(基于任务的自适应与规划多智能体系统),这是一个将大型语言模型(LLMs)与符号规划相结合的多智能体框架,旨在无需手动定义环境模型的情况下解决复杂任务。TAPAS采用专门的基于LLM的智能体,它们通过结构化的工具调用机制协作生成并适时调整领域模型、初始状态及目标设定。借助这种基于工具的交互方式,下游智能体可向上游智能体请求修改,从而适应新属性与约束,无需手动重新定义领域。结合自然语言计划翻译的ReAct(推理+行动)式执行智能体,有效弥合了动态生成计划与现实机器人能力之间的鸿沟。TAPAS在基准规划领域及VirtualHome模拟现实环境中均展现出卓越性能。
English
We introduce TAPAS (Task-based Adaptation and Planning using AgentS), a multi-agent framework that integrates Large Language Models (LLMs) with symbolic planning to solve complex tasks without the need for manually defined environment models. TAPAS employs specialized LLM-based agents that collaboratively generate and adapt domain models, initial states, and goal specifications as needed using structured tool-calling mechanisms. Through this tool-based interaction, downstream agents can request modifications from upstream agents, enabling adaptation to novel attributes and constraints without manual domain redefinition. A ReAct (Reason+Act)-style execution agent, coupled with natural language plan translation, bridges the gap between dynamically generated plans and real-world robot capabilities. TAPAS demonstrates strong performance in benchmark planning domains and in the VirtualHome simulated real-world environment.
PDF11June 30, 2025