RESEARCH PORTFOLIO
LLM Agent-Based Item Classification System
Developed a natural language-controlled item classification system by integrating an LLM agent with a custom-built Arduino-based wheel sorter.
Natural Language Command · Arduino Control · Logistics Device Design and Manufacturing
I developed a system that moves and classifies items based on the user's natural language commands. I configured the system to perform direction control, speed adjustment, and motor operation by connecting an LLM-based agent with Arduino, and demonstrated the item classification operation on a custom-built device.
Development Goal
I aimed to implement an item classification system where users can convey desired operations through natural language instead of directly inputting individual control commands for the device. To achieve this, I configured a control flow connecting user command interpretation to hardware execution.
Core Implementation
-
Natural language-based device control
I implemented an LLM agent to interpret user commands and connect them to device control functions such as direction, speed, and motor operation. -
Arduino-based actuator integration
I connected the agent's execution requests with Arduino-based hardware control, configuring the software commands to result in actual item movement. -
Manufacturing of item classification hardware
I directly manufactured a device for the item classification demonstration. I implemented the direction change operation of a wheel sorter and conducted movement and classification experiments utilizing real items.
Implementation Results
I built a system that connects natural language input, command interpretation, device control, and item movement. I confirmed the integration of the software and the actuator by demonstrating the classification operation on the actually manufactured hardware.