LLM-Enhanced metaheuristics with single-shot and few-shot learning for multi-robot exploration tasks

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F61989100%3A27240%2F25%3A10260595" target="_blank" >RIV/61989100:27240/25:10260595 - isvavai.cz</a>

Výsledek na webu

<a href="https://link.springer.com/article/10.1007/s10586-025-05585-6?utm_source=getftr&utm_medium=getftr&utm_campaign=getftr_pilot&getft_integrator=clarivate" target="_blank" >https://link.springer.com/article/10.1007/s10586-025-05585-6?utm_source=getftr&utm_medium=getftr&utm_campaign=getftr_pilot&getft_integrator=clarivate</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1007/s10586-025-05585-6" target="_blank" >10.1007/s10586-025-05585-6</a>

Jazyk výsledku

angličtina

Název v původním jazyce

LLM-Enhanced metaheuristics with single-shot and few-shot learning for multi-robot exploration tasks

Popis výsledku v původním jazyce

Multi-robot exploration in unknown environments is a challenging optimization task. Although many optimization algorithms exist, there has been limited exploration of Large Language Models (LLMs) in the literature to enhance these methods. This study investigates the use of LLMs to improve optimization strategies for robotic exploration considering two main approaches: zero-shot and few-shot learnings. In the former case, the LLM generates solutions based on a problem description alone. In the latter, few-shot learning, the LLM refines solutions based on initial results as well as a problem description. Using models like GPT, Gemini, and Claude, we created enhanced variants of Particle Swarm Optimization (PSO) to improve exploration efficiency and help robots navigate complex environments to address PSO’s common issues such as local optima. Experimental results demonstrate the strong potential of LLMs in refining optimization strategies for multi-robot exploration. © The Author(s) 2025.

Název v anglickém jazyce

LLM-Enhanced metaheuristics with single-shot and few-shot learning for multi-robot exploration tasks

Popis výsledku anglicky

Multi-robot exploration in unknown environments is a challenging optimization task. Although many optimization algorithms exist, there has been limited exploration of Large Language Models (LLMs) in the literature to enhance these methods. This study investigates the use of LLMs to improve optimization strategies for robotic exploration considering two main approaches: zero-shot and few-shot learnings. In the former case, the LLM generates solutions based on a problem description alone. In the latter, few-shot learning, the LLM refines solutions based on initial results as well as a problem description. Using models like GPT, Gemini, and Claude, we created enhanced variants of Particle Swarm Optimization (PSO) to improve exploration efficiency and help robots navigate complex environments to address PSO’s common issues such as local optima. Experimental results demonstrate the strong potential of LLMs in refining optimization strategies for multi-robot exploration. © The Author(s) 2025.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2025

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

Cluster Computing-The Journal of Networks Software Tools and Applications

ISSN

1386-7857

e-ISSN

—

Svazek periodika

28

Číslo periodika v rámci svazku

16

Stát vydavatele periodika

US - Spojené státy americké

Počet stran výsledku

35

Strana od-do

1-35

Kód UT WoS článku

001597096900032

EID výsledku v databázi Scopus

2-s2.0-105019198118