Scholarlink

This paper presents a novel method for plan generation with Large Language Models (LLMs). We propose utilizing Web-of-Thing Thing Descriptions (WoT TD) to inform the LLM about available devices for interaction. We investigate a novel pipeline in which: (1) The WoT TD gets translated into a Python class, (2) The task in natural language gets translated into code that interact with this class, (3) The generated code gets executed to obtain the plan. During execution we synchronize the state of the Python class with the state of the real device to enable dynamic adaptions of the plan. We evaluate our approach featuring 6 different LLMs. We get performance improvements of up to \(14\%\) , when comparing our approach against existing LLM-based planning methods. Furthermore, we explore the influence of different aspects of WoT TDs on planning capabilities. This research paves the way towards more potent LLM planning models by introducing Python classes as intermediaries, simplifying the final planning task while leveraging standardized and accessible domain knowledge provided by WoT TDs.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Static and Adaptive Planning with WoT TD by Generating Python Objects as Intermediary Representations Using Large Language Models

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Lukas Kinder,
  • Tobias Käfer

摘要

This paper presents a novel method for plan generation with Large Language Models (LLMs). We propose utilizing Web-of-Thing Thing Descriptions (WoT TD) to inform the LLM about available devices for interaction. We investigate a novel pipeline in which: (1) The WoT TD gets translated into a Python class, (2) The task in natural language gets translated into code that interact with this class, (3) The generated code gets executed to obtain the plan. During execution we synchronize the state of the Python class with the state of the real device to enable dynamic adaptions of the plan. We evaluate our approach featuring 6 different LLMs. We get performance improvements of up to \(14\%\) , when comparing our approach against existing LLM-based planning methods. Furthermore, we explore the influence of different aspects of WoT TDs on planning capabilities. This research paves the way towards more potent LLM planning models by introducing Python classes as intermediaries, simplifying the final planning task while leveraging standardized and accessible domain knowledge provided by WoT TDs.