Towards a next-generation LLM empowered low-code programming industrial robotic system for human-centric smart manufacturing

Abstract

Industrial robotic systems have been widely adopted in modern industries due to their advantages in high flexibility and strong adaptability. However, these systems are often limited by fragmented workflows, high cognitive demands on operators, and complex interaction programming. To address these issues, this study proposes a next-generation low-code programming framework empowered by large language models (LLMs), aiming to advance human-centric smart manufacturing (HCSM). By integrating the reasoning capabilities of LLMs into industrial robotic systems, the framework prioritizes intuitive, efficient, and operator-friendly interaction, establishing a novel paradigm for industrial applications. Additionally, the system incorporates a cognitive assistance module to reduce the cognitive burden on unskilled operators. Moreover, an LLM-based low-code programming module was designed, employing a multi-agent mechanism for intent recognition, parameter extraction, and human verification, thereby significantly enhancing the system's ability to robustly handle unstructured natural language instructions in industrial environments. Finally, the system was validated through a case study on aircraft panel drilling, demonstrating its practicality and reliability while supporting unskilled operators in performing complex tasks. This validation indicates that the proposed method has broad potential for industrial applications.