Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

LLM-Guided Decentralized Exploration with Self-Organizing Robot Teams

Hiroaki Kawashima, Shun Ikejima, Takeshi Takai, Mikita Miyaguchi, Yasuharu Kunii

TL;DR

This study proposes an exploration method that combines an algorithm for self-organization, enabling the autonomous and dynamic formation of multiple teams, and an algorithm that allows each team to autonomously determine its next exploration target (destination).

Abstract

When individual robots have limited sensing capabilities or insufficient fault tolerance, it becomes necessary for multiple robots to form teams during exploration, thereby increasing the collective observation range and reliability. Traditionally, swarm formation has often been managed by a central controller; however, from the perspectives of robustness and flexibility, it is preferable for the swarm to operate autonomously even in the absence of centralized control. In addition, the determination of exploration targets for each team is crucial for efficient exploration in such multi-team exploration scenarios. This study therefore proposes an exploration method that combines (1) an algorithm for self-organization, enabling the autonomous and dynamic formation of multiple teams, and (2) an algorithm that allows each team to autonomously determine its next exploration target (destination). In particular, for (2), this study explores a novel strategy based on large language models (LLMs), while classical frontier-based methods and deep reinforcement learning approaches have been widely studied. The effectiveness of the proposed method was validated through simulations involving tens to hundreds of robots.

LLM-Guided Decentralized Exploration with Self-Organizing Robot Teams

TL;DR

This study proposes an exploration method that combines an algorithm for self-organization, enabling the autonomous and dynamic formation of multiple teams, and an algorithm that allows each team to autonomously determine its next exploration target (destination).

Abstract

When individual robots have limited sensing capabilities or insufficient fault tolerance, it becomes necessary for multiple robots to form teams during exploration, thereby increasing the collective observation range and reliability. Traditionally, swarm formation has often been managed by a central controller; however, from the perspectives of robustness and flexibility, it is preferable for the swarm to operate autonomously even in the absence of centralized control. In addition, the determination of exploration targets for each team is crucial for efficient exploration in such multi-team exploration scenarios. This study therefore proposes an exploration method that combines (1) an algorithm for self-organization, enabling the autonomous and dynamic formation of multiple teams, and (2) an algorithm that allows each team to autonomously determine its next exploration target (destination). In particular, for (2), this study explores a novel strategy based on large language models (LLMs), while classical frontier-based methods and deep reinforcement learning approaches have been widely studied. The effectiveness of the proposed method was validated through simulations involving tens to hundreds of robots.
Paper Structure (13 sections, 8 figures)

This paper contains 13 sections, 8 figures.

Table of Contents

  1. Introduction
  2. Related work
  3. Decentralized exploration
  4. Robot model and environment representation
  5. Task description and team behavior
  6. Decentralized team formation
  7. LLM-based team destination determination
  8. Experiments
  9. Setup
  10. Self-organization of teams
  11. LLM-based destination selection
  12. Scalability with larger numbers of robots
  13. Conclusion

Figures (8)

  • Figure 1: Sensor model and occupancy grid map generation. In (b), occupancy probability values are indicated in grayscale: gray cells represent unexplored areas, white cells represent explored free areas (free space), and black cells represent explored areas with obstacles (occupied space). Green cells represent frontier cells.
  • Figure 2: Example of self-organized teams and a charging robot
  • Figure 3: Example exploration using the baseline method
  • Figure 4: Comparison of explored area between baseline and LLM-based methods (N = 15)
  • Figure 5: Example of LLM reasoning for destination selection
  • ...and 3 more figures