An End-to-end Planning Framework with Agentic LLMs and PDDL

2025-12-10

AI Summary

This work addresses the challenge of automatically and robustly translating natural language planning requirements into executable PDDL models. We propose the first LLM-driven, multi-agent collaborative end-to-end planning framework that achieves fully automated, closed-loop generation—from ambiguous or inconsistent natural language descriptions to syntactically correct and semantically valid PDDL domain and problem definitions—via semantic parsing, iterative multi-agent modeling, and formal verification. The framework integrates state-of-the-art planners (e.g., Fast Downward) and the VAL validator, supports temporal constraint modeling, and guarantees plan optimality where applicable. It further enables bidirectional translation between symbolic plans and natural language step sequences. Evaluated on NaturalPlan, PlanBench, and Blocksworld benchmarks, our approach significantly improves planning success rates, demonstrating exceptional robustness on small-scale complex reasoning tasks. To our knowledge, it is the first system achieving fully autonomous, human-intervention-free semantic-to-symbolic planning translation.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Multiagent Systems: Multiagent Planning
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

We present an end-to-end framework for planning supported by verifiers. An orchestrator receives a human specification written in natural language and converts it into a PDDL (Planning Domain Definition Language) model, where the domain and problem are iteratively refined by sub-modules (agents) to address common planning requirements, such as time constraints and optimality, as well as ambiguities and contradictions that may exist in the human specification. The validated domain and problem are then passed to an external planning engine to generate a plan. The orchestrator and agents are powered by Large Language Models (LLMs) and require no human intervention at any stage of the process. Finally, a module translates the final plan back into natural language to improve human readability while maintaining the correctness of each step. We demonstrate the flexibility and effectiveness of our framework across various domains and tasks, including the Google NaturalPlan benchmark and PlanBench, as well as planning problems like Blocksworld and the Tower of Hanoi (where LLMs are known to struggle even with small instances). Our framework can be integrated with any PDDL planning engine and validator (such as Fast Downward, LPG, POPF, VAL, and uVAL, which we have tested) and represents a significant step toward end-to-end planning aided by LLMs.

Problem

• Converts natural language specifications into PDDL models automatically
• Addresses ambiguities and contradictions in human specifications iteratively
• Generates and translates plans without human intervention using LLMs

Innovation

• Orchestrator converts natural language to PDDL
• LLM agents refine domain and problem automatically
• Framework integrates with external planning engines