Adaptive Domain Modeling with Language Models: A Multi-Agent Approach to Task Planning
This addresses the challenge of automating task planning for robots or AI systems in dynamic environments, though it appears incremental as it builds on existing multi-agent and LLM-based methods.
The paper tackles the problem of solving complex tasks without manually defined environment models by introducing TAPAS, a multi-agent framework that integrates Large Language Models with symbolic planning, resulting in strong performance in benchmark planning domains and the VirtualHome simulated environment.
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.