LLMPC: Large Language Model Predictive Control

2025-01-05

AI Summary

Large language models (LLMs) exhibit uncontrollable implicit optimization behaviors and suffer from inaccurate planning difficulty assessment in complex task planning. Method: We propose an explicit planning paradigm grounded in Model Predictive Control (MPC), establishing for the first time a theoretical correspondence between LLM-based planning and MPC. Our approach introduces a differentiable planning cost function, a multi-granularity action evaluator, and a cost-driven, plug-and-play mechanism that replaces conventional black-box prompt engineering. Integrated with structured reasoning prompts, it enables explicit, interpretable, and controllable planning. Contribution/Results: Evaluated on multiple complex task planning benchmarks, our method achieves a +12.7% improvement in task success rate and enhanced robustness. Empirical results demonstrate that cost guidance effectively calibrates LLMs’ implicit optimization processes, offering a novel, principled paradigm for controllable and interpretable LLM planning.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Recent advancements in prompting techniques for Large Language Models (LLMs) have improved their reasoning, planning, and action abilities. This paper examines these prompting techniques through the lens of model predictive control (MPC). We show that LLMs act as implicit planning cost function minimizers when planning prompts are used. Under our framework we demonstrate that LLM planning performance can be improved further by incorporating real planning cost functions and evaluators.

Problem

• Large Language Models
• Model Predictive Control
• Planning Optimization

Innovation

• Large Language Models
• Prompt Engineering
• Planning Performance Enhancement