Decompose, Plan in Parallel, and Merge: A Novel Paradigm for Large Language Models based Planning with Multiple Constraints

2025-06-03

AI Summary

To address constraint coupling and cascading errors in multi-constraint planning tasks for LLM-based agents, this paper proposes the Decompose–Parallel-Plan–Merge (DPPM) paradigm: constraint-aware task decomposition enables parallel subtask planning; a graph-structured merging mechanism and a conflict-aware verification module support dynamic error correction and reflective refinement. The core contributions are (i) the first multi-constraint-driven parallel planning framework, overcoming the error accumulation bottleneck inherent in sequential paradigms; and (ii) an end-to-end verifiable and correctable robust planning pipeline. Evaluated on a travel planning benchmark, DPPM achieves a 32.7% improvement in planning success rate, a 58.4% reduction in constraint violation rate, and a 41% decrease in average planning latency.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Multiagent Systems: Multiagent Planning
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Despite significant advances in Large Language Models (LLMs), planning tasks still present challenges for LLM-based agents. Existing planning methods face two key limitations: heavy constraints and cascading errors. To address these limitations, we propose a novel parallel planning paradigm, which Decomposes, Plans for subtasks in Parallel, and Merges subplans into a final plan (DPPM). Specifically, DPPM decomposes the complex task based on constraints into subtasks, generates the subplan for each subtask in parallel, and merges them into a global plan. In addition, our approach incorporates a verification and refinement module, enabling error correction and conflict resolution. Experimental results demonstrate that DPPM significantly outperforms existing methods in travel planning tasks.

Problem

• Addresses limitations in LLM-based planning tasks
• Proposes parallel planning paradigm for multiple constraints
• Improves performance in complex travel planning tasks

Innovation

• Decomposes complex tasks into subtasks
• Plans for subtasks in parallel
• Merges subplans into final plan