TY - CHAP

T1 - Paraphrasing Method for Controlling a Robotic Arm Using a Large Language Model

AU - Yasuda, Shohei

AU - Komori, Yuka

AU - Nakajima, Kosuke

AU - Tanabe, Ryota

AU - Uemura, Wataru

AU - Walter, Niclas

N1 - Conference code: 13

PY - 2024/11

Y1 - 2024/11

N2 - In recent years, factory production systems have shifted from low-mix high-volume production to high-mix low-volume production and the robots such as arm typed robots and mobile robots are installed in a factory. Integrating robots, sensors and the other devices into a system is costly. To solve the problems in regard to the costs, a method to automatically generate control code for a robot using a large language model (LLM) is considered. Controlling a robot using an LLM makes it possible for non experts in on-site adjustment of robots to adjust motion of a robot installed in factory by giving text-base instructions. In case to use an LLM, we must devise how to write instructions to obtain desired motion of the robot. In this paper, we address pick and place task using a robotic arm and propose a method to learn how to paraphrase instructions to high quality ones for desired controlling a robotic arm using an LLM in reinforcement learning.

AB - In recent years, factory production systems have shifted from low-mix high-volume production to high-mix low-volume production and the robots such as arm typed robots and mobile robots are installed in a factory. Integrating robots, sensors and the other devices into a system is costly. To solve the problems in regard to the costs, a method to automatically generate control code for a robot using a large language model (LLM) is considered. Controlling a robot using an LLM makes it possible for non experts in on-site adjustment of robots to adjust motion of a robot installed in factory by giving text-base instructions. In case to use an LLM, we must devise how to write instructions to obtain desired motion of the robot. In this paper, we address pick and place task using a robotic arm and propose a method to learn how to paraphrase instructions to high quality ones for desired controlling a robotic arm using an LLM in reinforcement learning.

KW - large language model

KW - paraphrase

KW - reinforcement learning

KW - robotic arm

KW - Engineering

UR - http://www.scopus.com/inward/record.url?scp=85213354505&partnerID=8YFLogxK

U2 - 10.1109/GCCE62371.2024.10760305

DO - 10.1109/GCCE62371.2024.10760305

M3 - Article in conference proceedings

AN - SCOPUS:85213354505

SN - 979-8-3503-5508-6

T3 - GCCE - IEEE Global Conference on Consumer Electronics

SP - 943

EP - 945

BT - GCCE 2024 - 2024 IEEE 13th Global Conference on Consumer Electronics

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 13th IEEE Global Conference on Consumer Electronic - GCCE 2024

Y2 - 29 October 2024 through 1 November 2024

ER -