//Dorian Lamouille
Back to projects
software

Research paper about LLM

Published and presented a research paper on IEEE for ARSO 2026 conference in Vienna : 'Benchmarking local language models for social robots using edge devices'

PythonRaspberryollamacuda
IEEE PaperarXiv Preprint

This is part of an open-source project called RSC (Robot Study Companion): https://rsc.ee. This project aims to create an affordable and open-source robot that can assist students and increase their performance.

As part of this project, I built from scratch a benchmark specialized to evaluate the performance of LLMs on edge devices and assess their teaching ability. I created this benchmark to find the best LLM for educational robots, then decided to publish those findings on IEEE and to participate in the 2026 ARSO conference in Vienna to present my research to other colleagues in the robotics field.