Understanding the planning of LLM agents: A survey

Notes from paper Understanding the planning of LLM agents: A survey by Xu Huang, Weiwen Liu, Xiaolong Chen, Xingmei Wang, Hao Wang, Defu Lian, Yasheng Wang, Ruiming Tang, Enhong Chen

Overview

A survey that explores the planning capabilities of Large Language Modelss used as AI Agents.

The authors categorise existing research into five directions: * Task Decomposition * Multi-Plan Selection * External Planner-Aided Planning * Reflection and Refinement * Memory-augmented Planning

Each section discusses the motivation, key approaches, and limitations of the respective direction.

The survey also presents a detailed analysis of LLM-based agents' planning abilities, highlighting challenges like: * Hallucinations * Feasibility of Plans * Efficiency of Plans * Multi-Modal Environmental Feedback