LLM+P: Empowering Large Language Models with Optimal Planning Proficiency

2023/04/22 by Bo Liu, Liu, Bo, Yuqian Jiang +11 · 1 voice · 129 citations
Computer Science · Mathematics · #Artificial intelligence #Benchmark (surveying) #Code (set theory) #Computer science #Generalization #Mathematics #Multimodal Machine Learning Applications #Natural Language Processing Techniques #Natural language #Operations research #Plan (archaeology) #Programming language #Set (abstract data type) #Topic Modeling

paper · pdf · doi:10.48550/arxiv.2304.11477

published in arXiv (Cornell University) (Cornell University)

openalex publication_date 2023/04/22 · openalex created_date 2023/04/27 · openalex updated_date 2026/08/03

Abstract

Large language models (LLMs) have demonstrated remarkable zero-shot generalization abilities: state-of-the-art chatbots can provide plausible answers to many common questions that arise in daily life. However, so far, LLMs cannot reliably solve long-horizon planning problems. By contrast, classical planners, once a problem is given in a formatted way, can use efficient search algorithms to quickly identify correct, or even optimal, plans. In an effort to get the best of both worlds, this paper introduces LLM+P, the first framework that incorporates the strengths of classical planners into LLMs. LLM+P takes in a natural language description of a planning problem, then returns a correct (or optimal) plan for solving that problem in natural language. LLM+P does so by first converting the language description into a file written in the planning domain definition language (PDDL), then leveraging classical planners to quickly find a solution, and then translating the found solution back into natural language. Along with LLM+P, we define a diverse set of different benchmark problems taken from common planning scenarios. Via a comprehensive set of experiments on these benchmark problems, we find that LLM+P is able to provide optimal solutions for most problems, while LLMs fail to provide even feasible plans for most problems.\footnoteThe code and results are publicly available at https://github.com/Cranial-XIX/llm-pddl.git.

Cited by

• Can LLMs Prove Robotic Path Planning Optimality? A Benchmark for Research-Level Algorithm Verification
• Learning Abstractions for Hierarchical Planning in Program-Synthesis Agents
• Embodied Robot Manipulation in the Era of Foundation Models: Planning and Learning Perspectives
• LEACL: LLM-Enhanced Automatic Curriculum Learning for Reinforcement Learning in Long-Horizon Manipulation Tasks
• SymStep: Symbolic Step Verification for Logical Reasoning
• HELP: Hierarchical Embodied Language Planner for Household Tasks
• LookPlanGraph: Embodied Instruction Following Method with VLM Graph Augmentation
• AnyTask: an Automated Task and Data Generation Framework for Advancing Sim-to-Real Policy Learning
• End-to-end PDDL Planning with Hardcoded and Dynamic Agents
• Principles2Plan: LLM-Guided System for Operationalising Ethical Principles into Plans
• The Vision Wormhole: Latent-Space Communication in Heterogeneous Multi-Agent Systems
• Geometrically-Constrained Agent for Spatial Reasoning
• Frontier Large Language Models Rival State-of-the-Art Planners
• HalluClean: A Unified Framework to Combat Hallucinations in LLMs
• LLM-GROP: Visually Grounded Robot Task and Motion Planning with Large Language Models
• Using Language Models as Closed-Loop High-Level Planners for Robotics Applications: A Brief Overview and Benchmarks
• CoFineLLM: Conformal Finetuning of LLMs for Language-Instructed Robot Planning
• Kinodynamic Task and Motion Planning using VLM-guided and Interleaved Sampling
• Simulation to Rules: A Dual-VLM Framework for Formal Visual Planning
• Large language model-based task planning for service robots: A review
• PIP-LLM: Integrating PDDL-Integer Programming with LLMs for Coordinating Multi-Robot Teams Using Natural Language
• Using Large Language Models for Abstraction of Planning Domains - Extended Version
• NeSyPr: Neurosymbolic Proceduralization For Efficient Embodied Reasoning
• Empowering Real-World: A Survey on the Technology, Practice, and Evaluation of LLM-driven Industry Agents
• T3 Planner: A Self-Correcting LLM Framework for Robotic Motion Planning with Temporal Logic
• Echoes of Human Malice in Agents: Benchmarking LLMs for Multi-Turn Online Harassment Attacks
• Adaptive Obstacle-Aware Task Assignment and Planning for Heterogeneous Robot Teaming
• DeepPlanner: Scaling Planning Capability for Deep Research Agents via Advantage Shaping
• A²FM: An Adaptive Agent Foundation Model for Tool-Aware Hybrid Reasoning
• ParaCook: On Time-Efficient Planning for Multi-Agent Systems
• Guiding Exploration in Reinforcement Learning Through LLM-Augmented Observations
• Agent+P: Guiding UI Agents via Symbolic Planning
• Language Model as Planner and Formalizer under Constraints
• Flash-Searcher: Fast and Effective Web Agents via DAG-Based Parallel Execution
• Alternatives To Next Token Prediction In Text Generation -- A Survey
• Explore-Execute Chain: Towards an Efficient Structured Reasoning Paradigm
• PRIME: Planning and Retrieval-Integrated Memory for Enhanced Reasoning
• Multi-Agent Path Finding via Offline RL and LLM Collaboration
• Goal-Guided Efficient Exploration via Large Language Model in Reinforcement Learning
• Self-CriTeach: LLM Self-Teaching and Self-Critiquing for Improving Robotic Planning via Automated Domain Generation
• ComposableNav: Instruction-Following Navigation in Dynamic Environments via Composable Diffusion
• LAD-VF: LLM-Automatic Differentiation Enables Fine-Tuning-Free Robot Planning from Formal Methods Feedback
• LLM-Guided Task- and Affordance-Level Exploration in Reinforcement Learning
• SPAR: Scalable LLM-based PDDL Domain Generation for Aerial Robotics
• Toward PDDL Planning Copilot
• AgenticIE: An Adaptive Agent for Information Extraction from Complex Regulatory Documents
• Teaching LLMs to Plan: Logical Chain-of-Thought Instruction Tuning for Symbolic Planning
• Virtual Agent Economies
• Robot guide with multi-agent control and automatic scenario generation with LLM
• AI Wellbeing
• Language Models For Generalised PDDL Planning: Synthesising Sound and Programmatic Policies
• Who Sees What? Structured Thought-Action Sequences for Epistemic Reasoning in LLMs
• Neuro-Symbolic Artificial Intelligence: Towards Improving the Reasoning Abilities of Large Language Models
• LOOP: A Plug-and-Play Neuro-Symbolic Framework for Enhancing Planning in Autonomous Systems
• Inspire or Predict? Exploring New Paradigms in Assisting Classical Planners with Large Language Models
• Large Model Empowered Embodied AI: A Survey on Decision-Making and Embodied Learning
• Symmetry-Aware Transformer Training for Automated Planning
• Triple-S: A Collaborative Multi-LLM Framework for Solving Long-Horizon Implicative Tasks in Robotics
• Constraint-Preserving Data Generation for Visuomotor Policy Learning
• HyCodePolicy: Hybrid Language Controllers for Multimodal Monitoring and Decision in Embodied Agents
• L3M+P: Lifelong Planning with Large Language Models
• How Far Are LLMs from Symbolic Planners? An NLP-Based Perspective
• Interleaved LLM and Motion Planning for Generalized Multi-Object Collection in Large Scene Graphs
• Can LLM-Reasoning Models Replace Classical Planning? A Benchmark Study
• Pretraining a Unified PDDL Domain from Real-World Demonstrations for Generalizable Robot Task Planning
• Enhancing Visual Planning with Auxiliary Tasks and Multi-token Prediction
• Theory of Mind Abilities of Large Language Models in Human-Robot Interaction: An Illusion?
• Conditional Multi-Stage Failure Recovery for Embodied Agents
• SEEA-R1: Tree-Structured Reinforcement Fine-Tuning for Self-Evolving Embodied Agents
• PSALM-V: Automating Symbolic Planning in Interactive Visual Environments with Large Language Models
• T-Rex: Task-Adaptive Spatial Representation Extraction for Robotic Manipulation with Vision-Language Models
• Adaptive Domain Modeling with Language Models: A Multi-Agent Approach to Task Planning
• Distilling On-device Language Models for Robot Planning with Minimal Human Intervention
• Language-Informed Synthesis of Rational Agent Models for Grounded Theory-of-Mind Reasoning On-The-Fly
• Context Matters! Relaxing Goals with LLMs for Feasible 3D Scene Planning
• Language-Grounded Hierarchical Planning and Execution with Multi-Robot 3D Scene Graphs
• Prime the search: Using large language models for guiding geometric task and motion planning by warm-starting tree search
• Enhance Multimodal Consistency and Coherence for Text-Image Plan Generation
• CP-Bench: Evaluating Large Language Models for Constraint Modelling
• Conversational Interfaces for Parametric Conceptual Architectural Design: Integrating Mixed Reality with LLM-driven Interaction
• Direct Behavior Optimization: Unlocking the Potential of Lightweight LLMs
• "Don't Do That!": Guiding Embodied Systems through Large Language Model-based Constraint Generation
• Grounded Vision-Language Interpreter for Long-Horizon Bimanual Task and Motion Planning
• Tru-POMDP: Task Planning Under Uncertainty via Tree of Hypotheses and Open-Ended POMDPs
• PGPO: Enhancing Agent Reasoning via Pseudocode-style Planning Guided Preference Optimization
• Earley-Driven Dynamic Pruning for Efficient Structured Decoding
• GridRoute: A Benchmark for LLM-Based Route Planning with Cardinal Movement in Grid Environments
• Automated Hybrid Reward Scheduling via Large Language Models for Robotic Skill Learning
• Learning Compositional Behaviors from Demonstration and Language
• Make Planning Research Rigorous Again!
• Large Language Models for Planning: A Comprehensive and Systematic Survey
• Robot Operation of Home Appliances by Reading User Manuals
• Subtle Risks, Critical Failures: A Framework for Diagnosing Physical Safety of LLMs for Embodied Decision Making
• Weights or Skills? A Survey of Robot-Learning Techniques: from Action-Predicting Weights to Robots that Write their Own Skills
• One Demo Is All It Takes: Planning Domain Derivation with LLMs from A Single Demonstration
• Global Optimization and Inference-Time Region Grafting for Agentic Workflows
• STRCMP: Integrating Graph Structural Priors with Language Models for Combinatorial Optimization
• Unifying Inference-Time Planning Language Generation
• Building a Stable Planner: An Extended Finite State Machine Based Planning Module for Mobile GUI Agent
• Are LLMs Better Formalizers than Solvers on Complex Problems?
• Zero-Shot Iterative Formalization and Planning in Partially Observable Environments
• ViPlan: A Benchmark for Visual Planning with Symbolic Predicates and Vision-Language Models
• RLAP: A Reinforcement Learning Enhanced Adaptive Planning Framework for Multi-step NLP Task Solving
• Learning Hierarchical Domain Models Through Environment-Grounded Interaction
• Air-Ground Collaboration for Language-Specified Missions in Unknown Environments
• Symbolic Attack Chain Generation from Atomic Red Team Techniques: An Empirical Study of Predicate Representation Granularity
• Strategy-Augmented Planning for Large Language Models via Opponent Exploitation
• Any House Any Task: Scalable Long-Horizon Planning for Abstract Human Tasks
• Semantic Intelligence: Integrating GPT-4 with A Planning in Low-Cost Robotics
• DUPLEX: Agentic Dual-System Planning via LLM-Driven Information Extraction
• From Woofs to Words: Towards Intelligent Robotic Guide Dogs with Verbal Communication
• Language Models Can Resolve Reference Compositionally, But It's Not Their Native Strength: The Case of the Personal Relation Task
• Leveraging Pre-trained Large Language Models with Refined Prompting for Online Task and Motion Planning
• AI Planning Framework for LLM-Based Web Agents
• The Long-Horizon Task Mirage? Diagnosing Where and Why Agentic Systems Break
• RAGEN: Understanding Self-Evolution in LLM Agents via Multi-Turn Reinforcement Learning
• Robo-Troj: Attacking LLM-based Task Planners
• Vision-Language Model Predictive Control for Manipulation Planning and Trajectory Generation
• Tree of Thoughts as a Classical Heuristic Search Problem: Formal Foundations and Design Patterns
• Hierarchical Planning for Complex Tasks with Knowledge Graph-RAG and Symbolic Verification
• Building LLM Agents by Incorporating Insights from Computer Systems
• Chain-of-Modality: Learning Manipulation Programs from Multimodal Human Videos with Vision-Language-Models
• Syntactic and Semantic Control of Large Language Models via Sequential Monte Carlo
• ADAPT: Actively Discovering and Adapting to Preferences for any Task
• InstructRAG: Leveraging Retrieval-Augmented Generation on Instruction Graphs for LLM-Based Task Planning
• Agentic Nesting: A New Methodology for Existing Enterprise Application Integration and Services
• On the Temporal Question-Answering Capabilities of Large Language Models Over Anonymized Data
• Hybrid Privilege Escalation and Remote Code Execution Exploit Chains
• Have Large Language Models Learned to Reason? A Characterization via 3-SAT Phase Transition

Discussions

• LLM with Planning [hn, 81 points, 16 comments]

Related

• Systematic Analysis of LLM Contributions to Planning: Solver, Verifier,\n Heuristic
• Exploring and Benchmarking the Planning Capabilities of Large Language Models
• Planning Anything with Rigor: General-Purpose Zero-Shot Planning with LLM-based Formalized Programming
• Inspire or Predict? Exploring New Paradigms in Assisting Classical Planners with Large Language Models
• Classical Planning with LLM-Generated Heuristics: Challenging the State of the Art with Python Code