A framework for reconfigurable production line changeover task planning based on large language model

Area of Science:

• Robotics and Automation
• Artificial Intelligence
• Manufacturing Systems Engineering

Background:

• Manual changeovers in reconfigurable production lines suffer from inefficiency and high error rates.
• Automating these operations is crucial for improving manufacturing flexibility and productivity.

Purpose of the Study:

• To propose an automatic changeover task planning framework for reconfigurable production lines.
• To leverage Large Language Models (LLMs) for interpreting user instructions and controlling robots.

Main Methods:

• A framework integrating a knowledge graph, high-level task planner, and low-level task planner.
• LLM-based interpretation of semantic instructions to decompose tasks and generate skill sequences.
• Skill-based modular programming with a four-layer control architecture.

Main Results:

• Demonstrated the framework's correctness through single-robot and actual production line experiments.
• Successfully controlled a fabricated changeover robot to perform automatic changeover tasks.
• Validated the framework's ability to interpret user instructions and execute automated changeovers.

Conclusions:

• The proposed LLM-based framework effectively addresses challenges in manual changeover operations.
• Enables automatic changeover tasks in reconfigurable production lines, enhancing efficiency and accuracy.
• Paves the way for more intelligent and autonomous manufacturing systems.