Fast and Accurate Task Planning using Neuro-Symbolic Language Models and Multi-level Goal Decomposition

2024-09-28

arXiv.org

AI Summary

To address the dual bottlenecks of low search efficiency in symbolic planners (e.g., PDDL) and poor success rates of LLM-based methods for long-horizon robotic task planning in complex environments, this paper proposes a neuro-symbolic fusion framework. It leverages LLMs for hierarchical goal decomposition and dynamically orchestrates either symbolic planners or MCTS-augmented LLM planners for each sub-goal. The framework introduces, for the first time, a dynamic sub-goal decomposition mechanism coupled with an adaptive planner coordination strategy—achieving both high inference speed and significantly improved accuracy. Moreover, it enables the first robust, end-to-end task planning with LLMs on real robots. Experiments across multiple benchmarks and real/simulated robotic platforms demonstrate a 57–73% reduction in planning latency and a task success rate of 91.4%, substantially outperforming both pure symbolic and pure LLM baselines.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Neuro-Symbolic Learning
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling

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

In robotic task planning, symbolic planners using rule-based representations like PDDL are effective but struggle with long-sequential tasks in complicated planning environments due to exponentially increasing search space. Recently, Large Language Models (LLMs) based on artificial neural networks have emerged as promising alternatives for autonomous robot task planning, offering faster inference and leveraging commonsense knowledge. However, they typically suffer from lower success rates. In this paper, to address the limitations of the current symbolic (slow speed) or LLM-based approaches (low accuracy), we propose a novel neuro-symbolic task planner that decomposes complex tasks into subgoals using LLM and carries out task planning for each subgoal using either symbolic or MCTS-based LLM planners, depending on the subgoal complexity. Generating subgoals helps reduce planning time and improve success rates by narrowing the overall search space and enabling LLMs to focus on smaller, more manageable tasks. Our method significantly reduces planning time while maintaining a competitive success rate, as demonstrated through experiments in different public task planning domains, as well as real-world and simulated robotics environments.

Problem

• Improve robotic task planning speed and accuracy
• Combine neuro-symbolic models for complex task decomposition
• Balance symbolic and LLM planners for subgoal efficiency

Innovation

• Neuro-symbolic task planner combining LLM and symbolic methods
• Multi-level goal decomposition to simplify complex tasks
• Dynamic subgoal planning with MCTS or symbolic planners