Hierarchical LLM-Based Multi-Agent Framework with Prompt Optimization for Multi-Robot Task Planning

Researchers developed a hierarchical multi-agent LLM framework that significantly improves multi-robot task planning by combining natural language processing with classical PDDL planners. The system uses prompt optimization and meta-learning to achieve success rates of up to 95% on compound tasks, outperforming previous state-of-the-art methods by substantial margins.

• →New hierarchical LLM framework combines language models with classical PDDL planners for multi-robot coordination
• →System achieves 95% success rate on compound tasks and 84% on complex tasks, significantly outperforming previous methods
• →TextGrad-inspired prompt optimization automatically improves planning accuracy when initial plans fail
• →Meta-prompts are shared across agents to enable efficient learning in multi-agent settings
• →Ablation study shows hierarchical structure contributes most (+59 points) to overall performance improvement